Louisiana State University LSU Digital Commons

LSU Doctoral Dissertations Graduate School

2011 Photochemical Reaction Products of Polycyclic Aromatic Hydrocarbons Adsorbed at an Air-Water Interface Franz Stefan Ehrenhauser Louisiana State University and Agricultural and Mechanical College, [email protected]

Follow this and additional works at: https://digitalcommons.lsu.edu/gradschool_dissertations Part of the Chemical Engineering Commons

Recommended Citation Ehrenhauser, Franz Stefan, "Photochemical Reaction Products of Polycyclic Aromatic Hydrocarbons Adsorbed at an Air-Water Interface" (2011). LSU Doctoral Dissertations. 530. https://digitalcommons.lsu.edu/gradschool_dissertations/530

This Dissertation is brought to you for free and open access by the Graduate School at LSU Digital Commons. It has been accepted for inclusion in LSU Doctoral Dissertations by an authorized graduate school editor of LSU Digital Commons. For more information, please [email protected]. PHOTOCHEMICAL REACTION PRODUCTS OF POLYCYCLIC AROMATIC HYDROCARBONS ADSORBED AT AN AIR-WATER INTERFACE

A Dissertation

Submitted to the Graduate Faculty of the Louisiana State University and Agricultural and Mechanical College in partial fulfillment of the requirements for the degree of Doctor of Philosophy in The Cain Department of Chemical Engineering

by Franz Stefan Ehrenhauser Dipl. Ing., Johannes Kepler University, Linz, Austria, 2009

December 2011

Die hohe Kraft Der Wissenschaft, Der ganzen Welt verborgen! Und wer nicht denkt, Dem wird sie geschenkt, Er hat sie ohne Sorgen.

Johann Wolfgang von Goethe, Faust, 1808

To the patience of my parents and my loving wife

ii ACKNOWLEDGEMENTS

I would like to thank first and foremost my advisor Dr. Mary J. Wornat for her support and her guidance. Thanks to her, doors and opportunities have opened for me, which I would have otherwise never realized. I also have to express my deepest gratitude to Dr. Kalliat T. Valsaraj, who acted as my co-advisor during these years, for his patience and continuous support. I would also like to thank my committee members, Dr. James J. Spivey, Dr. Francisco Hung and Dr.

Maud Walsh for their time.

I have to thank all my colleagues and co-workers for their help, their inspirations through fruitful discussions and their additional perspectives. I want to especially thank Dr. Jing Chen, to whom I am greatly indebted to, as without her work the extent of this dissertation would not have been possible. I also owe gratitude to all the members of the department of chemical engineering, which took care, that either ideas could be realized in the laboratory (Paul Rodriguez, Joe Bell,

Fred McKenzie), or that the bureaucracy did not stand in the way (Darla Dao, Melanie

McCandless and Danny Fontenot).

I want to thank my parents for their patience and their continuous support. A very special thanks bears to my beloved wife Daira, who stood by me these years, being my big support in any difficult situation.

I would like to acknowledge funding from NSF (Grant ATM 0355291 and Grant ATM

0907261), the Air Force Office of Scientific Research for providing funding for two HPLC instruments (DURIP Grant FA9550-05-1-0253 and DURIP Grant FA9550-08-1-0281), the LSU

Graduate School for providing additional scholarships, as well Dr. Charles Coates, whose generosity in form of the Dr. Charles E. Coates Scholar Research Award allowed me to collect new experiences and present my work worldwide.

iii TABLE OF CONTENTS

ACKNOWLEDGEMENTS ...... iii

LIST OF TABLES ...... vii

LIST OF FIGURES ...... viii

ABSTRACT ...... xiv

1. INTRODUCTION ...... 1 1.1 Polycyclic Aromatic Hydrocarbons ...... 2 1.2. PAH Photodegradation ...... 5 1.2.1. Degradation Products of PAH in Bulk Phases ...... 9 1.2.2. Degradation Products of PAH in Thin Water Films ...... 10 1.3. Environmental Impacts of Volatile PAH and Their Oxidation Products ...... 12 1.3.1. Health Effects of Volatile PAH and Their Oxidation Products ...... 12 1.3.2. Environmental Effects of Volatile PAH and Oxidation Products ...... 13 1.4. Objective and Structure of Thesis ...... 14

2. ANALYSIS OF UV-OXIDATION PRODUCTS OF PAH – METHOD DEVELOPMENT ...... 15 2.1 HPLC – Method Development ...... 17 2.2. Online Sample Concentration ...... 21 2.3. UV Absorption as Detection Method for Oxygenated Polycyclic Aromatic Compounds ...... 25 2.4. APPI-MS ...... 29 2.5. APPI-MS Method Development ...... 31 2.5.1. APPI Source Optimization ...... 33 2.5.2. Dopant Delivery System ...... 39 2.5.2.1. Design of the Dopant Delivery System ...... 40 2.5.2.2. Dopant Selection ...... 45 2.5.2.3. Ionization Performance of Benzene as Dopant ...... 49 2.6. APPI-MS Spectra ...... 53 2.6.1. Polycyclic Aromatic Hydrocarbons ...... 54 2.6.2. Hydroxyl-Substituted Polycyclic Aromatic Hydrocarbons ...... 55 2.6.3. Alcohols ...... 56 2.6.4. Aromatic Ketones ...... 57 2.6.5. Carboxylic Acids and Esters ...... 58 2.6.6. Aldehydes ...... 60 2.6.7. APPI-MS Spectra for the Identification of OPAC ...... 61

3. EXPERIMENTAL SETUP ...... 64 3.1. Thin-Film Reactor ...... 64 3.2. Bulk-Phase Reactor ...... 69 3.3. Sample Analysis ...... 70

iv 4. PHOTOOXIDATION PRODUCTS OF POLYCYCLIC AROMATIC HYDROCARBONS ...... 72 4.1. Naphthalene ...... 72 4.2. Phenanthrene ...... 74 4.3. Pyrene ...... 81 4.4. Acenaphthene ...... 83 4.5. 9H-Fluorene ...... 85 4.6. Summary ...... 93

5. PHOTOOXIDATION OF 9H-FLUORENE IN THIN WATER FILMS ...... 95 5.1. Uptake of 9H-Fluorene onto Water Films ...... 95 5.2 Photooxidation of 9H-Fluorene in Thin-Water Films ...... 98 5.3. Discussion ...... 103

6. CONCLUSIONS...... 107 6.1. Advances in the Analytical Methodology of OPAC...... 107 6.2. Newly Identified UV Oxidation Products...... 108 6.3. Interfacial Degradation of 9H-Fluorene ...... 109 6.4. Suggested Future Work ...... 109

REFERENCES ...... 111

APPENDIX A: HPLC METHODS ...... 122

APPENDIX B: GLASS SLIDE PREPARATION ...... 128

APPENDIX C: PURIFICATION OF POLYCYCLIC AROMATIC HYDROCARBONS ...... 132 Naphthalene ...... 132 Phenanthrene ...... 132 9H-Fluorene ...... 134 Acenaphthene ...... 135 Pyrene ...... 135

APPENDIX D: UV SPECTRAL MATCHES ...... 136

APPENDIX E: SYNTHETIC PROCEDURES ...... 140 Synthesis of 6H-Dibenzo[b,d]pyran-6-one ...... 141 Synthesis of 6H-Dibenzo[b,d]pyran-6-ol...... 143 Synthesis of 2-(2’-Hydroxy-phenyl)-benzyl alcohol ...... 145 Synthesis of 2- and 3- Hydroxy-9,10-phenanthrenequinone ...... 148 Synthesis of 2- and 3-Phenanthrene Sulfonates ...... 149 Synthesis of 2-Hydroxyphenanthrene ...... 150 Synthesis of 2-Acetoxyphenanthrene and 2-Acetoxy-9,10-phenanthrenequinone ...... 152 Synthesis of 2-Hydroxy-9,10-phenanthrenequinone ...... 153 Synthesis of 3-Hydroxyphenanthrene ...... 155 Synthesis of 3-Acetoxyphenanthrene ...... 157 Synthesis of 3-Acetoxy-9,10-phenanthrenequinone ...... 158 Synthesis of 3-Hydroxy-9,10-phenanthrenequinone ...... 159

v Synthesis of 1-Acenaphthenone ...... 161

APPENDIX F: PERMISSION TO REPRINT ...... 163

VITA ...... 164

vi LIST OF TABLES

Table 1. Properties of 16 PAH ...... 4

Table 2. Relative potency factors of some PAH18 ...... 4

Table 3. First-order kinetic constants for 9H-fluorene degradation in pure water ...... 10

Table 4. Photodegradation products of volatile PAH in pure water (bulk) ...... 11

Table 5. LOD and recoveries of PAH and OPAC with a 5-mL injection volume with online concentration, compared to traditional liquid-liquid extraction with GC-MS analysis ...... 25

Table 6. Ionization mechanisms104,108-116 in APPI mass spectrometry...... 32

Table 7. Optimized APPI-MS parameters for qualitative analysis at 0.2 mL/min mobile phase flow (methanol) ...... 53

Table 8. Observed APPI-MS ionization patterns of polycyclic aromatic compounds..... 62

Table 9. Kinetic constants for 9H-fluorene oxidation products formed in a 185-µm water film and a 37-µm water film at 19.8°C ...... 102

Table 10. Mask material for HF etching ...... 129

vii LIST OF FIGURES

Figure 1. 16 polycyclic aromatic hydrocarbons as EPA priority pollutants: I naphthalene, II acenaphthylene, III acenaphthene, IV 9H-fluorene, V anthracene, VI phenanthrene, VII fluoranthene, VIII pyrene, IX benzo[k]fluoranthene, X benzo[b]fluoranthene, XI chrysene, XII benz[a]anthracene, XIII benzo[a]pyrene, XIV benzo[ghi]perylene, XV dibenz[a,h]anthracene, XVI indeno[1,2,3-cd]pyrene ...... 3

Figure 2. Emission spectrum of sunlight at air mass 1.5 (AM 1.5 ground-level), UVB-Lamp (UVP Inc., Cambridge, UK) and absorbance spectra of volatile PAH...... 6

Figure 3. Generalized reaction pathways of PAH degradation with singlet oxygen, with hydroxyl 1 3 radicals and by direct ionization ( O2 singlet oxygen, O2 triplet oxygen) ...... 7

Figure 4. Riedl Pfleiderer process for the commercial production of hydrogen peroxide ...... 8

Figure 5. Self condensation of anthracene adsorbed on silica ...... 9

Figure 6. HPLC chromatogram of naphthalene photooxidation mixture analyzed with PAH standard method (Method A in Appendix A)...... 18

Figure 7. HPLC chromatogram of naphthalene photooxidation mixture analyzed on a Pinnacle II Phenyl column (Restek Corp., Bellefonte, PA, USA) ...... 20

Figure 8. HPLC chromatogram of naphthalene photooxidation mixture analyzed on a Ultra Aqueous C18 column (Restek Corp. Bellefonte, PA, USA) ...... 21

Figure 9. Six-port switching valve configuration; for online micro solid-phase extraction system a) loading the large volume sample into the sample loop, b) loading the sample onto the guard column c) switching the guard column into the analytical flow path ...... 23

Figure 10. UV absorbance spectra of pyrene (black), 4,5-pyrenequinone (green), 1,8- pyrenequinone (blue) and 1,6-pyrenequinone (red) ...... 27

Figure 11. UV absorbance spectra of naphthalic acid anhydride (red), naphthalic acid (black) and naphthalic acid dianion (blue) ...... 28

Figure 12. Two commercially available ionization sources: (a) linear PhotoSpray® and (b) orthogonal PhotoMate® ...... 33

Figure 13. Naphthalene (m/z 128) ionization dependency on the nebulizer pressure ...... 35

Figure 14. Naphthalene (m/z 128) ionization dependency on the vaporizer temperature ...... 36

Figure 15. Efficiency of naphthalene ionization as a function of drying gas flow and drying gas temperature...... 37

viii Figure 16. Peak area obtained from extracted positive ion chromatograms [M-2 to M+2] of polycyclic aromatic compounds versus the capillary voltage...... 38

Figure 17. The effect of the capillary voltage on the ionization pattern of 9,10- phenanthrenequinone without a dopant...... 39

Figure 18. Schematic of the gas-phase dopant delivery system ...... 42

Figure 19. Benzene delivery rate at 40°C saturator temperature...... 46

Figure 20. Positive (a) and negative (b) ionization efficiencies with anisole, acetone, benzene and toluene as dopants; for the negative ions of 1-naphthoic acid m/z 125-129 was used instead of m/z 170-174 ...... 48

Figure 21. Peak area obtained from extracted positive ion chromatogram [M-2 to M+2] of polycyclic aromatic compounds versus the capillary voltage. Benzene was deployed as a dopant at 40 mg/min...... 50

Figure 22. The effect of the capillary voltage on the ionization pattern of 9,10- phenanthrenequinone [PHQ] with 40 mg/min benzene as dopant...... 51

Figure 23. APPI-MS spectra of 9H-fluorene (a) and phenanthrene (b) with benzene as dopant; the red line marks the abundance of the positive and the blue line the abundance of the negative ions ...... 55

Figure 24. APPI-MS spectra of phenol (a) and naphthalene-1-ol (b) with benzene as dopant; the red line marks the abundance of the positive and the blue line the abundance of the negative ions ...... 56

Figure 25. APPI-MS spectra of 1-indanol (a) and 9-hydroxy-9H-fluorene (b) with benzene as dopant; the red line marks the abundance of the positive and the blue line the abundance of the negative ions ...... 57

Figure 26. APPI-MS spectra of 9H-fluorene-9-one (a) and 9,10-phenanthrenequinone (b) with benzene as dopant; the red line marks the abundance of the positive and the blue line the abundance of the negative ions ...... 58

Figure 27. APPI-MS spectra of 1-naphthoic acid (a) and 6H-dibenzo[b,d]pyran-6-one (b) with benzene as dopant; the red line marks the abundance of the positive and the blue line the abundance of the negative ions ...... 59

Figure 28. APPI-MS spectra of trans-cinnamaldehyde (a) and 1-naphthaldehyde (b) with benzene as dopant; the red line marks the abundance of the positive and the blue line the abundance of the negative ions ...... 60

Figure 29. Thin-film reactor (schematic). Solid arrows represent gas flow; dashed lines represent coolant flow ...... 66

ix Figure 30. Cross-sectional view of the temperature-controlled boat reactor ...... 66

Figure 31. Cross-sectional view of the bulk-phase photoreactor ...... 70

Figure 32. HPLC-chromatogram of naphthalene photooxidation mixture from the UVB photooxidation of naphthalene adsorbed and dissolved in 450-µm thick-water film ...... 73

Figure 33. HPLC-chromatogram of naphthalene photooxidation mixture from the UVB photooxidation of naphthalene adsorbed and dissolved in 22-µm thick water film ...... 74

Figure 34. HPLC chromatogram from a photooxidation mixture of phenanthrene obtained from UVB irradiation of a phenanthrene-saturated 515-µm thick-water film ...... 75

Figure 35. HPLC chromatogram of a photooxidation mixture of very pure phenanthrene (>99.5%) obtained from 12 hours UVB irradiation of a saturated phenanthrene solution in the bulk-phase reactor...... 76

Figure 36. a) UV spectrum of unknown A and b) mass spectrum of unknown A ...... 77

Figure 37. a) APPI-MS mass spectrum of synthesized 2-hydroxy-9,10-phenanthrenequinone and b) UV spectra of unknown A and synthesized 2-hydroxy-9,10-phenanthrenequinone ...... 80

Figure 38. HPLC chromatogram of the photooxidation mixture of a half-saturated pyrene mixture after 12 hours UVB irradiation...... 82

Figure 39. UV spectra of 4,5-pyrenequinone and pyrene photooxidation product identified as 4,5-pyrenequinone ...... 83

Figure 40. HPLC chromatogram of the photooxidation mixture of an aqueous, half-saturated acenaphthene solution after 12 hours UVB irradiation, 500-µL injection volume ...... 84

Figure 41. HPLC chromatogram from a photooxidation mixture of 9H-fluorene obtained from UVB irradiation of an aqueous, half-saturated 9H-fluorene 369-µm thick water film ...... 86

Figure 42. UV absorbance spectrum of unknown B ...... 87

Figure 43. a) positive and b) negative APPI-MS mass spectrum of unknown B ...... 88

Figure 44. Potential candidates for unknown B, containing two phenyl rings, one carbonyl and one hydroxyl group, with elemental composition C13H10O2...... 89

Figure 45. Ionization mechanism of 6H-dibenzo[b,d]pyran-6-ol during electron-impact ionization...... 90

Figure 46. Suggested ionization mechanism for 6H-dibenzo[b,d]pyran-6-ol in dopant-assisted APPI-MS. Negative ions in blue; positive ions in red ...... 91

Figure 47. a) Mass spectrum of synthesized 6H-dibenzo[b,d]pyran-6-ol and b) UV spectra of unknown B and synthesized 6H-dibenzo[b,d]pyran-6-ol ...... 92

x Figure 48. Thermodynamic equilibrium relationships between bulk air, bulk water and the air-water interface ...... 96

Figure 49. Concentration of 9H-fluorene-9-ol, 6H-dibenzo[b,d]pyran-6-ol, 9H-fluorene-9-one and 9H-fluorene in the 185-µm film during photodegradation (lines fitted to Equation 11) ...... 99

Figure 50. Concentration of 9H-fluorene-9-ol, 6H-dibenzo[b,d]pyran-6-ol, 9H-fluorene-9-one and 9H-fluorene in the 37-µm film during photodegradation (lines fitted to Equation 11) ...... 100

Figure 51. Possible reaction pathways for 9H-fluorene photooxidation ...... 105

Figure 52. a) UV absorbance spectra of a naphthalene photooxidation product (black line) and phtalide (red line), and b) UV absorbance spectra of a naphthalene photooxidation product (black line) and coumarin (red line)...... 136

Figure 53. a) UV absorbance spectra of a naphthalene photooxidation product (black line) and 1- indanone (red line), and b) UV absorbance spectra of a naphthalene photooxidation product (black line) and 1,4-naphthoquinone (red line)...... 136

Figure 54. UV absorbance spectra of a naphthalene photooxidation product (black line) and naphthalene-1-ol (red line)...... 137

Figure 55. a) UV absorbance spectra of a phenanthrene photooxidation product (black line) and 9H-fluorene (red line), and b) UV absorbance spectra of a phenanthrene photooxidation product (black line) and 9H-fluorene-9-one (red line)...... 137

Figure 56. a) UV absorbance spectra of a phenanthrene photooxidation product (black line) and 6H-dibenzo[b,d]pyran-6-one (red line), and b) UV absorbance spectra of a phenanthrene photooxidation product (black line) and 9,10-phenanthrenequinone (red line)...... 137

Figure 57. a) UV absorbance spectra of a acenaphthene photooxidation product (black line) and 1-acenaphthenone (red line), and b) UV absorbance spectra of a acenaphthene photooxidation product (black line) and 1-acenaphthenol (red line)...... 138

Figure 58. a) UV absorbance spectra of a pyrene photooxidation product (black line) and 1,8- pyrenequinone (red line) obtained from the literature,177 and b) UV absorbance spectra of a pyrene photooxidation product (black line) and 1,6-pyrenequinone (red line)...... 138

Figure 59. UV absorbance spectra of a pyrene photooxidation product (black line) and 1- hydroxypyrene (red line)...... 139

Figure 60. UV absorbance spectra of a 9H-fluorene photooxidation product (black line) and 9H-fluorene-9-ol (red line) ...... 139

Figure 61. Bayer-Villiger Oxidation of 9H-fluorene-9-one to 6H-dibenzo[b,d]pyran-6-one according to Mehta et al.164 ...... 141

xi 1 Figure 62. 400 MHz H-NMR spectrum of 6H-dibenzo[b,d]pyran-6-one in CDCl3 ...... 142

Figure 63. UV absorbance spectrum and mass spectrum of 6H-dibenzo[b,d]pyran-6-one...... 143

Figure 64. Synthesis of 6H-dibenzo[b,d]pyran-6-ol following the procedure of Moliner et al.160 ...... 143

1 Figure 65. 400 MHz H-NMR spectrum of crude 6H-dibenzo[b,d]pyran-6-one in CDCl3...... 144

Figure 66. UV absorbance spectrum and mass spectrum of 6H-dibenzo[b,d]pyran-6-ol ...... 145

Figure 67. Reduction of 6H-dibenzo[b,d]pyran-6-one with sodium borohydride resulting in 2-(2’-hydroxy-phenyl) benzyl alcohol ...... 145

Figure 68. 400 MHz 1H-NMR spectrum of 2-(2’-hydroxy-phenyl)-benzyl alcohol in CDCl3 ...... 147

Figure 69. UV absorbance spectrum and mass spectrum of 2-(2’-hydroxyphenyl)-benzyl alcohol ...... 148

Figure 70. Synthetic scheme for the synthesis of 2- and 3-hydroxy-9,10- phenanthrenequinone according to Fieser180 and Werner et al.148-149 ...... 148

Figure 71. Synthesis of 2- and 3-phenanthrene sulfonate according to Fieser180 ...... 149

Figure 72. Transformation of barium 2-phenanthrenesulfonate into 2-hydroxyphenanthrene181 ...... 150

1 Figure 73. 400 MHz H-NMR spectrum of 2-hydroxyphenanthrene in CDCl3 ...... 151

Figure 74. UV absorbance spectrum and mass spectrum of 2-hydroxyphenanthrene ...... 152

Figure 75. Acetylation of 2-hydroxyphenanthrene and oxidation182 to 2-acetoxy-9,10- phenanthrenequinone ...... 152

Figure 76. Hydrolysis of 2-acetoxy-9,10-phenanthrenequinone to 2-hydroxyphenanthrene- quinone ...... 153

Figure 77. 400 MHz 1H-NMR spectrum of 2-hydroxy-9,10-phenanthrenequinone in DMSO-d6 ...... 154

Figure 78. UV absorbance spectrum and mass spectrum of 2-hydroxy-9,10- phenanthrenequinone ...... 155

Figure 79. Transformation of potassium phenanthrene-3-sulfonate into 3- hydroxyphenanthrene181 ...... 155

1 Figure 80. 400 MHz H-NMR spectrum of 3-hydroxyphenanthrene in CDCl3 ...... 156

xii Figure 81. UV absorbance spectrum and mass spectrum of 3-hydroxyphenanthrene ...... 156

Figure 82. Acetylation of 3-hydroxyphenanthrene to 3-acetoxyphenanthrene ...... 157

Figure 83. Oxidation of 3-acetoxyphenanthrene to 3-acetoxyphenanthrene-9,10-quinone with chromium trioxide182...... 158

Figure 84. Hydrolysis of 3-acetoxy-9,10-phenanthrenequinone to 3- hydroxyphenanthrenequinone ...... 159

Figure 85. 400 MHz 1H-NMR spectrum of 3-hydroxyphenanthrene-9,10-quinone in DMSO-d6 ...... 160

Figure 86. UV absorbance spectrum and mass spectrum of 3-hydroxyphenanthrene-9,10- quinone ...... 160

Figure 87. Oxidation of acenaphthene with chromium trioxide ...... 161

1 Figure 88. 400 MHz H-NMR spectrum of 1-acenaphthenone in CDCl3 ...... 162

Figure 89. UV absorbance spectrum and mass spectrum of 1-acenaphthenone ...... 162

xiii ABSTRACT

Polycyclic aromatic hydrocarbons (PAH) are ubiquitous pollutants resulting from incomplete combustion and hence normally unavoidable. The kinetic aspects of the environmental degradation process have received considerable attention; however, not much is known about the degradation products themselves.

Aqueous solutions of naphthalene, acenaphthene, 9H-fluorene, phenanthrene and pyrene were subjected to UVB light in a bulk-phase and a thin-film reactor to assess the identity of the formed photodegradation products. To aide in the challenging analysis of these mixtures, a novel analytical technique based on high performance liquid chromatography (HPLC) coupled to UV diode-array detection and dopant-assisted atmospheric pressure photoionization mass spectrometry (APPI-MS) was developed, including the optimization of the chromatography, the introduction of an online-concentration system for increasing sensitivity and the deployment of a novel gas-phase dopant delivery system for APPI-MS. The dopant-assisted APPI-MS (DAPPI-

MS) was further optimized, to allow for qualitative analysis of oxygenated polycyclic aromatic compounds (OPAC). The application of the developed analytical technique allowed for the identification of 2-hdyroxy-9,10-phenanthrenequinone and 6H-dibenzo[b,d]pyran-6-ol as degradation products of phenanthrene and 9H-fluorene for the first time.

Most studies address PAH degradation in the bulk-water phase, neglecting the fact that the most significant contribution to their environmental degradation takes place in aerosols, especially for volatile PAH (2-4 aromatic rings). PAH can accumulate on the air-water interface, which provides a unique venue for their photochemical degradation. The interfacial photodegradation of 9H-fluorene was evaluated by measuring the formation rate of its photooxidation products in films of different thickness in the thin-film reactor. By increasing the

xiv surface-to-volume ratio of a thin film by a factor of five, the formation rate of 9H-fluorene-9-ol,

9H-fluorene-9-one and 6H-dibenzo[b,d]pyran-6-ol was significantly enhanced, highlighting the importance of interfacial degradation of organic pollutants on hydrometeors with large surface- to-volume ratio, such as fog.

xv 1. INTRODUCTION

Air-water interfaces are the most abundant interfaces on the planet. Besides the fact that two-thirds of the earth is covered with water, the weather and all biological systems rely on water as a primary medium. Important environmental surfaces are clouds, fog, open water, ocean spray, condensing industrial steam, snow, and any other hydrometeor. Fog, which provides ample surface, is of special interest, as it develops close to emission sources and human habitation.

In the 1950s, the pollutant-loaded form of fog, smog, dominated the media, and it has caused the American (and also international) legislature to enact more rigid environmental protective laws.1 Though its frequency of occurrence is reducing in North America and Europe due to the enacted clean air laws, it is still recurring in heavily populated areas with industry and during heavy traffic times. Furthermore, it plays and will play an even more significant role with the increasing motorization and industrialization of developing and newly industrialized countries such as India and China.

Smog can be a direct cause for the increase in fatalities and also cause long term effects such as bronchitis and lung cancer.2 It is also known that asthma cases are statistically higher in heavily polluted areas and that air pollution exacerbates existing asthma and heart diseases.3

Even though not directly toxic, just simple fog also has a substantial contribution to economic cost due to higher traffic accident rates, cancelled/delayed flights and rerouting in the trucking industry. The National Oceanic and Atmospheric Administration (NOAA) estimates fog and ice to account for up to 29 million USD in additional cost for the trucking industry. Weather-related delays in 2007 made up 65% of all delays in US air traffic, causing a 4.2 billion USD loss in economic efficiencies.4-5

1 Fog and subsequently smog formation are facilitated by small particles (e.g. PM2.5 and

PM10, particulate matter up to 2.5 and 10 µm respectively), by gaseous pollutants such as sulfur oxides and also by hydrocarbons and their degradation products.6 Fog and other hydrometeors also provide a unique venue for chemical transformation, the air-water interface. The air-water interface provides a different environment for the transformation of organic pollutants such as polycyclic aromatic hydrocarbons (PAH), compared to the bulk gas and liquid phase.

1.1 Polycyclic Aromatic Hydrocarbons

Polycyclic aromatic hydrocarbons (PAH) are the most abundant class of organic compounds in the universe7-9 and are linked to several important processes, potentially even the creation of life itself.10-11 However, they are mostly known as carcinogenic pollutants, especially benzo[a]pyrene as one of the most potent compounds. Polycyclic aromatic hydrocarbons consist of two or more rings formed of hydrogen and carbon, where at least one ring is of aromatic structure. Because of the molecules’ specific electronic configuration, they experience increased stability and chemical behavior that is significantly different from that of unsaturated, non- aromatic compounds. PAH can form from various processes, with the most abundant formation pathways being pyrolysis and incomplete combustion of carbon-based fuels.12 The formation can be quite complex and is influenced by a myriad of factors, always yielding complex mixtures of

PAH, which can be emitted into the atmosphere. As an illustrative example, the pyrolysis of the one-ring molecule 1,2-dihydroxybenzene (catechol), a model compound for low-rank coal and biomass, yields upon pyrolysis more than 70 individual PAH.13-16

As PAH are prevalent as pollutants of our combustion-driven society, the EPA defines 16 priority pollutants from the family of PAH,17 which act as representatives for the very complex

2 mixtures of polycyclic organic matter. Figure 1 shows the structural formulae of the 16 individual compounds, with their names given in the caption.

Figure 1. 16 polycyclic aromatic hydrocarbons as EPA priority pollutants: I naphthalene, II acenaphthylene, III acenaphthene, IV 9H-fluorene, V anthracene, VI phenanthrene, VII fluoranthene, VIII pyrene, IX benzo[k]fluoranthene, X benzo[b]fluoranthene, XI chrysene, XII benz[a]anthracene, XIII benzo[a]pyrene, XIV benzo[ghi]perylene, XV dibenz[a,h]anthracene, XVI indeno[1,2,3-cd]pyrene

Table 1 summarizes some basic values regarding the physical characterization of these priority pollutants. These 16 PAH are hydrophobic hydrocarbons, seven of them carcinogenic

(IX-XII, XV-XVI).18 Their water solubility is quite low, and several of them are considered semi-volatiles. PAH containing up to four rings can be found predominantly in the gas phase— specially naphthalene, 9H-fluorene, acenaphthylene, acenaphthene, anthracene, phenanthrene and pyrene (compounds I to VIII in Figure 1)—and are called ―volatile‖ PAH in this dissertation.

Larger molecules (compounds VIII to XVI in Figure 1) with significantly lower vapor pressures are predominantly found associated with particulate matter.19 The 16 EPA PAH are somewhat arbitrarily chosen, and there are several other PAH that should be considered because of their greater inherent carcinogenicity. Table 2 shows some of these PAH, such as dibenzo[a,l]pyrene,

3 Table 1. Properties of 16 PAH

PAH Water Chemical Enthalpy of Henry’s Law Vapor solubility formula sublimation constant 25 ºC pressure 25 ºC [g m-3] [kJ/mol] [L Pa mol-1×10-3] † [Pa]

Naphthalene 31.0 C10H8 72 50.1 10.4

Acenaphthylene 16.1 C12H8 73.2 12.3 0.9

Acenaphthene 3.8 C12H10 86 14.79 0.3

9H-Fluorene 1.9 C13H10 83 9.33 0.09

Anthracene 0.045 C14H10 98 5.01 0.0010

Phenanthrene 4.57 C14H10 91 4.46 0.020

Fluoranthene 0.26 C16H10 98 1.44 0.00123

Pyrene 0.132 C16H10 99 1.31 0.0006

-5 Benz[a]anthracene 0.0006 C18H12 109 0.691 2.8×10

* -7 Chrysene 0.0018 C18H12 118 0.447 5.7×10

-8 Benzo[k]fluoranthene 0.0008 C20H12 130 0.0589 5.2×10

-6† Benzo[b]fluoranthene 0.0015 C20H12 n/a 0.0661 7.5×10

-7 Benzo[a]pyrene 0.0038 C20H12 118.3 0.0463 7×10

-8 Benzo[ghi]perylene 0.00026 C22H12 127 0.0339 1.4×10

* -10 Indeno[1,2,3-cd]pyrene 0.00019 C22H12 n/a 0.0355 1.0×10

-10 Dibenz[a,h]anthracene 0.0006 C22H14 141.8 n/a 3.7×10

Values from Dabestani et al.20 except * from Gómez-Gutiérrez et al.21 and † from Ma et al.22 dibenzo[a,i]pyrene, dibenzo[a,h]pyrene, which are not part of the EPA 16 PAH but show significant carcinogenicity compared to the reference compound benzo[a]pyrene. The selection of the EPA 16 PAH also neglects the fact that alkyl substitution (e.g. methyl) can affect the carcinogenic property significantly, as can be seen in Table 2 by comparing the relative carcinogenic potency factors of chrysene and 5-methylchrysene. Nonetheless, the EPA 16 PAH

4 are significant, as they serve their purpose, to represent the incredibly complex mixtures encountered in the environment, very well.

Table 2. Relative potency factors of some PAH18

Compound Relative Potency

Benzo[a]pyrene 1

Dibenzo[a,h]pyrene 10

Dibenzo[a,i]pyrene 10

Dibenzo[a,l]pyrene 10

Chrysene 0.01

5-Methylchrysene 1

Benz[a]anthracene 0.1

7,12-Dimethylbenz[a]anthracene 6.4

1.2. PAH Photodegradation

Most PAH are chemically quite stable due to their aromatic character; however, as they show significant light absorbance in the ultraviolet B (UVB) and C (UVC) ranges, they are susceptible to photo-initiated degradation. Figure 2 shows the UV-absorbance spectra of the volatile PAH in comparison with the emission spectrum of natural sunlight and the commercial

UVB lamp, used in this work. As can be seen in Figure 2, natural sunlight provides electromagnetic waves with wavelengths of 250 to 2500 nm. Especially the range of 280 to

315 nm (UVB) is of interest for photodegradation, due to its high energy content (4.43-3.94 eV), its presence in natural sunlight, and as all volatile PAH of the EPA PAH 16 absorb in this range,

5 as demonstrated in Figure 2. The ultraviolet (UV) absorption of the volatile PAH allows them to interact as active reactants in the environment.

Refererence AM 1.5 Spectra Spectral Output ASTM G173/03 Visible UVB Lamp (UVP inc.) Visible UVC UVB UVA UVC UVB UVA

Intensity [arbitrary units]

250 275 300 325 350 375 400 425 450 250 275 300 325 350 375 400 425 450 Wavelength [nm] Wavelength [nm]

Figure 2. Emission spectrum of sunlight at air mass 1.5 (AM 1.5 ground-level), UVB-Lamp (UVP Inc., Cambridge, UK) and absorbance spectra of volatile PAH.

PAH also undergo reaction with active atmospheric species such as hydroxyl radicals,23 singlet oxygen24 and ozone.25 Figure 3 shows a schematic of the most common reaction pathways of PAH in the atmosphere. A major reaction route is via reaction with singlet oxygen

(the green reaction path in Figure 3). This metastable form of elemental oxygen is more reactive than the ground state triplet oxygen present in air and water and is generated by reversing the spin of one valence electron. There are two known ways to accomplish this: chemically (e.g. in the hypochlorite-hydrogen-peroxide reaction) and through photosensitization, by using a dye molecule such as a PAH as photocatalyst. Singlet oxygen is known to undergo [2+4] cycloaddition, as well as epoxy group formation with polycyclic aromatic hydrocarbons.24 PAH

6 can also react directly with the elemental oxygen (black reaction pathway in Figure 3) via an intermediate complex formation; however, this reaction path is of less significance in the environment.

The hydroxyl radical (blue reaction pathway in Figure 3) can form through reactions of ozone with water, the photolytic scission of hydrogen peroxide and reactions of NOx, nitrites and nitrates in the atmosphere.26-29 These processes are especially of importance in the troposphere and therefore concern cloud and fog chemistry. The hydroxyl radical is considered the

―detergent‖ of the troposphere; however, its contribution to ground-based reactions is mainly linked to ozone, as the reaction of ozone with water is the primary source for the radical.30

Figure 3. Generalized reaction pathways of PAH degradation with singlet oxygen, with hydroxyl 1 3 radicals and by direct ionization ( O2 singlet oxygen, O2 triplet oxygen)

Direct ionization, the red ionization pathway in Figure 3, seems at first glance unlikely, as the ionization potentials of PAH lie in the range of 6-8 eV, well below the energy range of terrestrial solar influx (<4.43eV or >280 nm). However, studies by Gudipati et al. and Woon et

7 al. have shown that solubilization in water or ice can significantly reduce this potential.31-32

Another indication for the possibility of direct ionization is reported by Ibrahim et al., who showed that benzene, as the simplest aromatic hydrocarbon, can form clusters with water, which potentially facilitate its ionization.33

Nearly every PAH shows photosensitization of oxygen (part of the green reaction pathway in Figure 3).20 The PAH gets electronically excited by collision with a photon and generates, by transferring energy to a triplet oxygen, a more reactive species, the singlet oxygen.

Photocatalytic behavior towards species other than oxygen cannot be excluded, but has not been widely investigated at this point. An example of an oxygenated polycyclic aromatic hydrocarbon acting as catalyst is the commercial generation of hydrogen peroxide in the Riedl Pfleiderer

Process, whose reaction pathway is shown in Figure 4.34 9,10-anthraquinone is first reduced with hydrogen to 9,10-anthrahydroquinone, which upon oxidation with air simultaneously forms hydrogen peroxide and regenerates 9,10-anthraquinone. The initial reduction with hydrogen in

Figure 4 can also be a photoreduction. Carapellucci et al. have shown that 9,10- phenanthrenequinone can undergo a photocatalytic cycle like anthraquinone, and oxidize other species.35 As such, photocatalytic behavior is a possible capability for any quinone-forming

PAH.

Figure 4. Riedl Pfleiderer process for the commercial production of hydrogen peroxide

Another potential reaction pathway in the phototransformation of PAH is the condensation of PAH or the condensation of oxygenated products. Figure 5 shows the

8 condensation products of anthracene, as they occur during UV-photooxidation of anthracene adsorbed on silica.36 This type of condensation of volatile organic matter is of considerable interest, as it could potentially explain some of the relatively large species found in fog waters,37 which make up a large portion of the organic matter.

Figure 5. Self condensation of anthracene adsorbed on silica

In the environment, the combination of these reaction pathways—reaction with (singlet) oxygen or hydroxyl radical, direct ionization, photosensitization and self-condensation—can get quite complex and intertwined. For instance, anthracene, one of the most reactive PAH, forms anthraquinone38 even without ultraviolet light and upon UV irradiation can facilitate oxidation of otherwise photostable products.39-40

1.2.1. Degradation Products of PAH in Bulk Phases

There have been many kinetic studies of PAH degradation in bulk water.39,41-58

Unfortunately there exists some divergence in the literature concerning reaction rates, partly due to very different light sources and/or reaction conditions. As an illustrative example, first-order reaction rate constants for the photochemical decomposition of 9H-fluorene in pure water are given in Table 3.

Although 9H-fluorene degradation in pure water is a comparatively simple process, with

9H-fluorene-9-one as the primary product, the wide range of values in Table 3 reveal that there is a controversy in the literature about degradation rates. Even more astounding is the scarcity of

9 Table 3. First-order kinetic constants for 9H-fluorene degradation in pure water

k [min-1]

Jacek and Dorota 43 0.099

Wai et al. 45 0.026

Sabaté et al.49 0.35

Trapido et al.59 0.105

literature about the actual products formed from the photodegradation. Even the simplest PAH, naphthalene, decomposes to more than 50 compounds as found by McConkey et al., but only eleven have been identified.39 Nonetheless, many oxidation products of PAH photodegradation are known. Table 4 represents a short summary of reported oxidation products of volatile PAH, obtained by photolysis in pure water. Particularly interesting is the fact that there are compounds, such as fluoranthene and acenaphthene, which have not been investigated in terms of their photodegradation in pure water. Except for a few compounds such as anthracene, a complete description of all products formed prior to the formation of photochemical stable compounds is missing.

1.2.2. Degradation Products of PAH in Thin Water Films

The focus of the majority of investigations found in the literature lies primarily on the reaction rate of the pollutants in the bulk phase or adsorbed on solid particles.49-56,68

Liquid/gaseous interfacial degradation has been investigated only rarely69-73 and most of the time without considering the formed products, whereas for bulk-phase reaction (gas or liquid phase) several pathways are known.46-48 Especially investigations of the air/water interface might reveal

10 Table 4. Photodegradation products of volatile PAH in pure water (bulk)

PAH Photolysis products

Naphthalene39 1-hydroxynaphthalene, 1,2-naphthoquinone, 1,4-naphthoquinone, phthalic acid, phthalic acid anhydride, 8-hydroxy-1,4- naphthoquinone, (3H)-2-hydroxy-1,4-naphthoquinone, coumarin, phthalide, cis-cinnamaldehyde

Acenaphthylene 57,60 1,2-acenaphthenequinone, naphthalene-1,8-dicarboxylic anhydride, acenaphthene, 1-hydroxyacenaphthene, naphthalene- 1,8-carboxaldehyde, naphthalene-1,8-carboxylic acid, cis/trans- acenaphth[1,2-a]acenaphthylene

Acenaphthene*61 1-acenaphthenone, 1-hydroxyacenaphthene, acenaphthylene

9H-Fluorene62-63 9H-fluorene-9-one, 9-hydroxy-9H-fluorene, 2-hydroxy-9H- fluorene

Anthracene41,58,64 9,10-anthraquinone, 1-hydroxy-9,10-anthraquinone, 2-hydroxy- 9,10-anthraquinone, 1-hydroxy-9,10-anthraquinone, 1,2- dihydroxy-9,10-anthraquinone, 1,3-dihydroxy-9,10-anthraquinone, 1,4-dihydroxy-9,10-anthraquinone, 1,5-dihydroxy-9,10- anthraquinone, 1,8-dihydroxy-9,10-anthraquinone, 2,6-dihydroxy- 9,10-anthraquinone, 1,2,4-trihydroxy-9,10-anthraquinone, 1,2,5,8- tetrahydroxy-9,10-anthraquinone, 1,2,10-trihydroxy-anthracene, phenol, benzaldehyde, salicylic aldehyde, phthalic acid, 2,5- dihydroxy benzoic acid, 2,5-dihydroxy benzoic aldehyde, benzoic acid, 1,2-benzene dialdehyde, 2-formyl benzoic acid, 2-hydroxy benzaldehyde, catechol, 2-hydroxy-1,4-naphthoquinone

Phenanthrene65-66 9,10-phenanthrenequinone, 9-hydroxy-phenanthrene, 9,10-epoxy- 9,10-dihydrophenanthrene, dibenz[b,d]oxepin, fluorene, 9H- fluorene-9-one, 6H-dibenzo[b,d]pyran-6-one, diphenic acid, diphenic acid anhydride, phthalic acid anhydride, 2,2’-diformyl biphenyl, 2-formylbiphenyl-2’-carboxylic acid

Fluoranthene**67 Adsorbed on glass: 9H-fluorene-9-one-1-carboxylic acid, 1,8- naphthalic anhydride, benzoic acid, phthalic acid, 1,2,4-benzene tricarboxylic acid, maleic acid, succinic acid, malonic acid, oxalic acid, naphthalene-1,8-dicarboxylic acid, 9H-fluorene-9-one, hydroxynaphthalic acid, 2-(1-hydroxyethyl)indene-1,3-dione, hydroxy-9H-fluorene-9-one

Pyrene42 1-hydroxypyrene, 1,6-pyrenequinone, 1,8-pyrenequinone * adsorbed on silica, ** adsorbed on glass

11 additional pathways for the destruction of PAH, as air/water interfaces have been shown to sustain chemistries74-76 substantially different from these in bulk phases.

PAH are interesting pollutants in conjunction with hydrometeors such as fog, as Herckes et al.77 have shown that polycyclic aromatic hydrocarbons can be found in significant concentrations in fog water, even surpassing their actual bulk water solubility. The enhanced concentration are due to an accumulation of PAH at the interface.71,73,78-80 Molecular simulations indicate that the interfacial region provides an energetic sink for the enrichment of PAH.81

Though the interfacial region is very thin, only a few Å thick,75,82 it seems reasonable to attribute it significant importance as venue for the photodegradation of PAH, as the highest PAH concentration, as well as the highest intensity of light incur at this position. Current research indicates that the air/water interface38,70-71,73,83 as well as the air/ice interface70,84-85 enhance the

PAH photodegradation rates.

1.3. Environmental Impacts of Volatile PAH and Their Oxidation Products

The environmental impacts of volatile PAH and their oxidation products can be split into two major areas: first, human health-related issues, where PAH and oxygenated polycyclic aromatic compounds (OPAC) might play significant roles, and second, their potential impact on atmospheric phenomena.

1.3.1. Health Effects of Volatile PAH and Their Oxidation Products

Although most of the volatile PAH are not carcinogenic, they are suspected to be at least chemical irritants. Several of the more polar and hence more bioavailable oxidation products are toxic or their impact is not known. The carcinogenicity of PAH stems from covalent bonding of

12 PAH metabolites with DNA and ultimately inducing breaking of DNA strands.86-87 This damage is the cause for cancer as well as for mutations. The primary metabolic transformation for carcinogenic PAH is usually an epoxidation step followed by hydrolysis to a diol and eventual additional epoxidation. The epoxides can bind covalently to a DNA strand and cause the disruption. These epoxides also form during environmental degradation with singlet oxygen and can contribute to the known phototoxicity of PAH.88 The direct increased health risk of photolysis products of PAH becomes evident in the case of naphthalene

89 (LD50orl,rat 490 mg/kg;) and its oxidation products 1,4-naphthoquinone (LD50orl,rat 190

90 91 mg/kg) and 1,2-naphthoquinone (LDLoorl,rat 250mg/kg). Toxicity measured as LD50orl,rat

(50% kill rate of rats by ingestion) increases by a factor of 2.5 for 1,4-naphthoquinone compared to naphthalene. A similar concentration of ingested 1,2-naphthoquinone would be already deadly. (LDLo – lowest known lethal dose) As such, even minor reaction products are of interest, as the increased toxicity can outweigh their lesser prevalence.

1.3.2. Environmental Effects of Volatile PAH and Oxidation Products

As PAH undergo reactive transformations to more polar and therefore more hydrophilic compounds, they are suspected to play a role in fog formation.92 The oxidation products of PAH, as polar and ambiphilic compounds, can contribute towards the stabilization of fog droplets and act as nuclei for the actual formation of fog.93-94

Photochemical smog is another atmospheric event that can be affected by PAH and

OPAC. Photochemical smog is a special form of smog, which forms upon sunlight interaction

95 with air pollution to form reactive species like NOx, ozone and peroxyacetyl nitrate.

Hydrocarbons have been shown to facilitate the formation of photochemical smog.6 Polycyclic

13 aromatic hydrocarbons, with their UV absorbance and their capability to increase the presence of oxidative species such as singlet oxygen, might facilitate the formation of photochemical smog.

1.4. Objective and Structure of Thesis

PAH present an important class of organic pollutants with significant effects on human life and economic cost. Although their role as carcinogens is well established, the role of their degradation products is practically unknown, as the products themselves are not well known. It seems imperative to establish more knowledge about PAH degradation and the degradation products, especially for the air-water interface. This work intends to enhance the understanding of the photochemical transformation of polycyclic aromatic hydrocarbons in and on thin water films, by assessing their photodegradation in a thin-film reactor, emulating large-surface-to- volume-ratio venues as they are encountered in hydrometeors such as fog.

This work is organized into two parts, the first part (Chapters 2, 3 and 4) describes the development of a novel analytical procedure for the analysis of oxidation products of PAH, as well as the application of the developed analytical techniques to the identification of the UV- oxidation products of naphthalene, acenaphthene, phenanthrene, 9H-fluorene and pyrene. As the evaluation of the photooxidation products of 9H-fluorene was particularly successful, Chapter 5 reports the evaluation of 9H-fluorene’s photooxidation in thin water films.

14 2. ANALYSIS OF UV-OXIDATION PRODUCTS OF PAH – METHOD DEVELOPMENT

The chemical analysis of polycyclic aromatic hydrocarbons, although they are molecules composed of only two elements, is a challenging task, as the number of possible isomers for these compounds runs into very large numbers. For just basic benzenoid PAH, which consist only of 6-membered rings, there is only one two-ring isomer, naphthalene. For benzenoid PAH with three rings, there are two unsubstituted isomers, phenanthrene and anthracene, and eight possible, single methyl-substituted derivatives. For dimethylated three-ring compounds, a total of

41 isomers can be obtained. By expanding this already complicated picture with another element, oxygen, it is evident that the complexity concerning the identification of individual oxygenated polycyclic aromatic compounds can be rather challenging.

As such, any experimental evaluation of the oxidation of PAH will center on the choice of proper model substances, such as the 16 EPA PAH, as a real world sample cannot be completely assessed with currently available technology. By starting with a single model compound, the complexity is greatly reduced, but nonetheless for example for 9H-fluorene, a three-ring non-benzenoid compound, five hydroxyfluorenes are possible, and twenty dihydroxy derivatives are possible. As oxygen is the basis for a variety of functional groups (alcohols, phenols, carboxyaldehyde, carboxylic acids, carboxylic ester, semi-acetales, acetales, ketals, ethers), the situation with oxygenated polycyclic aromatic compounds, is indeed very complicated.

The analysis of organic compounds in complex mixtures incorporates usually an advanced separation technique such as chromatography, coupled with an online detection method. For the chromatographic analysis of polycyclic aromatic compounds there are three

15 basic techniques: gas chromatography (GC), high performance liquid chromatography (HPLC) and micellar electrokinetic chromatography (MEKC, a form of capillary electrophoresis). Out of these techniques, HPLC analysis is particularly useful for oxygenated polycyclic aromatic compounds (OPAC):

 as it is a fairly simple method (compared to MEKC),

 as the increased polarity, low volatility and thermal instability of oxygenated

polycyclic aromatic compounds can be a problematic issue for GC analysis

 and as OPAC are easily detectable with typical HPLC detectors such as UV

absorption.

HPLC also faces some drawbacks such as the lower separation efficiency (lower plate number) compared to GC or MEKC and the more complex interfacing with mass spectrometric detectors, which are invaluable for qualitative identifications. Nonetheless, it is the most promising method for the qualitative and quantitative analysis of OPAC.

In the following, the development of a suitable analytical HPLC method for the identification and quantification of OPAC with UV diode-array detection and atmospheric pressure photoionization (APPI) mass spectrometric detection is presented. In order to assess complex mixtures of photooxidation products of PAH qualitatively and quantitatively, the chromatographic separation of such mixtures was tested with different HPLC columns and optimized. In addition, an advanced concentration system, based on micro solid-phase extraction was developed and tested. To enable mass spectral detection of OPAC, an APPI source was enhanced with a gas-phase dopant delivery system. The mass spectral information for different compound classes obtained with the dopant enhanced APPI mass spectrometer was evaluated and is discussed in the following.

16 2.1 HPLC – Method Development

HPLC is an advanced separation technique based on the different phase distribution between a liquid mobile phase and a solid stationary phase. The different interactions of compounds with the two phases allows the separation of compounds in mixtures, depending on the choice of mobile phase and stationary phase, according to their size, shape, polarity or number of π-bonds. The most commonly employed HPLC methods are based on reversed-phase chromatography with a nonpolar stationary phase (such as octadecyl surface-modified silica particles) and a polar mobile phase (water, acetonitrile, methanol).

The HPLC analysis of oxidation products of polycyclic aromatic hydrocarbons involves several specific challenges, as the range of polarity (one of the key parameters for reversed-phase chromatographic separation) of the individual compounds is wide, encompassing less-polar ketones to very polar carboxylic acids. Additionally the concentration is expected to be very low, as the solubility of the parent-PAH is at best in the mg/L-range, hence yielding PAH-oxidation products starting at the µg/L to mg/L range.

As an initial test for the analysis, a standard method for PAH (Method A in Appendix A) was evaluated,15 which employs a gradient elution program starting at 60:40 (vol/vol) water:acetonitrile, ramping up to 100% acetonitrile within 40 min, and then changing gradually to 100% dichloromethane within 40 min. Figure 6 shows the chromatogram of a naphthalene photooxidation sample (saturated naphthalene water solution, 60 min UVB irradiation, room temperature). In Figure 6, it can be seen that the standard analysis is clearly inadequate to assess oxygenated species, as any potential products elute before the naphthalene peak without ever experiencing sufficient separation. It is also evident that the entire chromatogram is too long and the portion of interest is very short. Therefore, the development of a new chromatographic

17 method was needed for the analysis of UV-oxidation products of polycyclic aromatic hydrocarbons. The primary challenge rested within the basic task of enhancing the retention of oxygenated organic species, so that the entire chromatogram can be utilized for separation.

100

80

60

base line rise due to 40 UV-Oxidation Products UV-absorbing solvent

20

Average UV Absorbance 190-520 nm [mAU] 190-520 UV Absorbance Average 0 0 20 40 60 80 HPLC Retention Time [min]

Figure 6. HPLC chromatogram of naphthalene photooxidation mixture analyzed with PAH standard method (Method A in Appendix A).

The chromatographic retention of compounds on the stationary phase (column) can be enhanced in two ways, either by decreasing the solvent strength of the mobile phase or by enhancing the adsorptive capacity of the stationary phase. With the polymeric-bound octadecyl silica phase used in the standard PAH method (Vydac 501TP54, 4.6×250 mm, W. R. Grace &

Co.-Conn., Columbia, MD, USA), reducing the solvent strength did not accomplish the goal of enhanced separation, as the use of a highly aqueous mobile phase caused the ―dewetting‖ of the

18 stationary phase,96 which impaired the reproducibility of any chromatographic separation.

Therefore, a two-pronged strategy was deployed, first testing the retention of the naphthalene photooxidation mixture with a phenyl-substituted stationary phase (Pinnacle II Phenyl, Restek

Corp. Bellefonte, PA, USA), which enhances the retention capacity of the stationary phase by targeting the aromatic body of oxygenated polycyclic aromatic compounds, and second by utilizing a special octadecyl phase (Ultra Aqueous C18, Restek Corp, Bellefonte, PA, USA), which is capable of sustaining reproducible separation under 100% aqueous conditions and therefore allows the use of an eluting solvent with a lower strength. Simultaneously, the change from 4.6-mm-diameter columns, used in the standard method, to narrow-bore 2.1-mm columns, allows an increase in sensitivity, as the injection-volume-to-flow-rate ratio is nearly doubled

(from 20 µL per 1.5 mL/min to 5 µL per 0.2 mL/min).

Figure 7 shows the chromatogram of the naphthalene photooxidation mixture analyzed with a 2.1×250 mm Pinnacle II Phenyl column (Restek Corp., Bellefonte, PA, USA) with a simple gradient method (Method B Appendix A) starting at 50% acetonitrile (rest water) with a hold of 30 min and ramping up to 100% acetonitrile within 30 min and a 30 min hold at this composition. Due to the reversed-phase character of the column the mobile phase strength could not be adjusted below 30% organic modifier (acetonitrile or methanol) to avoid dewetting. From a comparison of the chromatograms of the standard PAH method and the analysis on the phenyl- substituted column, it is clear that both columns yield similar retention behavior of the tested naphthalene-photooxidation sample, even though the underlying principles of interactions are different for the two columns (standard method–hydrophobicity; phenyl column– aromatic/aromatic π-bond interaction). As such, the tested phenyl column did not deliver a

19 suitable solution as the separation is nearly identical to the standard method, and lowering of the mobile phase strength is not possible due to the dewetting of the column.

350 UV-Oxidation Products

300

250

200

150

100

50

Average UV Absorbance 250-258 nm [mAU] 250-258 UV Absorbance Average 0

0 20 40 60 80

HPLC Retention Time [min]

Figure 7. HPLC chromatogram of naphthalene photooxidation mixture analyzed on a Pinnacle II Phenyl column (Restek Corp., Bellefonte, PA, USA)

The second strategy of lowering the solvent strength seemed to be therefore more promising. Figure 8 shows the chromatogram of the naphthalene photooxidation mixture analyzed with an 2.1×250 mm Ultra Aqueous C18 Column (Restek Corp., Bellefonte, PA, USA).

This column is specifically designed to avoid the effect of dewetting and therefore suitable for aqueous mobile phases (with water being the solvent with the lowest elution strength in reversed- phase chromatography). The chromatogram in Figure 8 was obtained by running a gradient elution initiating at 100% water (with a 30 min initial hold) within 60 min to 100% methanol

20 (with 30 min hold). As can be clearly seen in Figure 8, the column spreads all the components well through the entire chromatogram, enabling separation and ultimately their individual detection and identification. The initial solvent program has been further optimized for the identification of oxygenated compounds by testing with reference compounds, and a universal chromatographic method for the analysis of oxygenated polycyclic aromatic compounds resulting from the UV photooxidation of PAH was obtained (Method C Appendix A).

160

140

120

100

80 UV-Oxidation Products

60

40

20

0

Average UV Absorbance 250-258 nm [mAU] 250-258 UV Absorbance Average

0 20 40 60 80 HPLC Retention Time [min]

Figure 8. HPLC chromatogram of naphthalene photooxidation mixture analyzed on a Ultra Aqueous C18 column (Restek Corp. Bellefonte, PA, USA)

2.2. Online Sample Concentration

The analysis of oxidation products of volatile PAH faces another significant challenge.

As the aqueous solubility of the volatile PAH is low, ranging from 31 mg/L (naphthalene) to

21 0.132 mg/L (pyrene), the concentration of photooxidation products will be below these concentrations. To obtain a usable UV and/or mass spectrum, approximately 1 ng of the substance on-column (2.1-mm narrow-bore column) is necessary, enabling an observable concentration of 0.2 mg/mL for a typical 5 µL injection. At 0.2 mg/mL, naphthalene’s oxidation products can be observed easily; however, in the case of pyrene, it would be difficult to asses the oxidation, as the oxidation products would be two orders of magnitude lower in concentration.

An increase of the injection volume to overcome this problem, would mean a substantial decrease in chromatographic separation performance. Therefore, a concentration step is necessary.

The classical way of concentrating and determining organic compounds in water incorporates the extraction of the compounds with a water-immiscible volatile solvent such as dichloromethane. After concentrating the volatile solvent, the mixture is analyzed by GC or solvent-exchanged into an HPLC-compatible solvent such as acetonitrile or dimethylsulfoxide for HPLC analysis. This procedure is tedious, time-consuming and suffers from limited recovery of especially very polar compounds. Additionally, due to the thermal stress during the evaporation and the relative long handling time of 2-3 hours per sample, sample integrity cannot be guaranteed for the rather sensitive photooxidation products.

A more modern approach is solid-phase extraction, where a solid adsorbent extracts the organic matter from the aqueous phase. The organic compounds are back-extracted with a smaller volume of a stronger solvent such as methanol, enabling a concentration step. The concentrated solution is then again analyzed via HPLC or GC.

For this work a more direct approach was developed. As the HPLC method development had shown, the tested Ultra Aqueous material is highly retentive for organic compounds, even in

22 a 100% aqueous mobile phase. Instead of back-extracting the organic compounds from the solid extraction phase, the adsorbent is brought directly into the analytical flow-path, acting as injection and therefore reducing the time and solvent consumption significantly. By using a small

(2.1×10 mm) Ultra Aqueous C18 guard column as extraction cartridge, it is possible to inject large (mL) amounts of sample onto the analytical column, without sacrificing separation performance.

a) Guard b) Waste Column

Sample

Sample Loop Pump Waste Buffer

Six-Port Injection Valve Six-Port Switching Valve

Pump c) HPLC

Analytical Column

Figure 9. Six-port switching valve configuration; for online micro solid-phase extraction system a) loading the large volume sample into the sample loop, b) loading the sample onto the guard column c) switching the guard column into the analytical flow path

Although the extraction could be done offline, it is certainly more convenient to have a system online with the analytical chromatographic system. The devised system consists of two six-port valves, an additional HPLC pump, a large volume sample loop (500 µL or 5 mL) and a small 2.1x10 mm Ultra Aqueous C18 guard column, which acts as solid-phase extraction cartridge. Figure 9 shows the configuration of the two six-port valves with the sample loop, the

23 guard column and the flow path. During the initial step, depicted in Figure 9a, a large sample volume is loaded into the sample loop (up to 5 mL) and is then injected onto the guard column with pure water or buffer as eluting solvent, as shown in Figure 9b. As the Ultra Aqueous column material is highly retentive for organic compounds, it adsorbs practically all the organic analytes onto the very small volume (ca. 25 µL) of the guard column. The guard column is then brought into the analytical flow path (shown in Figure 9c) and acts as injection loop for the analysis, increasing the limits of detection (LOD) by up to three orders of magnitude. This method allows therefore for a very quick online concentration of aqueous samples with excellent recoveries and limits of detection.

The LOD and the recovery of the newly developed micro solid-phase-extraction method were compared with the classical approach of dichloromethane extraction, concentration in a

Kuderna-Danish apparatus and GC-MS analysis. Table 5 shows the values of the LOD for the online concentration of several compounds and compares them with the results obtained from the classical GC-MS analysis. Recoveries are above 85%, and LOD values are three orders of magnitude better than for an equivalent dichloromethane extraction/GC-MS analysis. As such the online micro solid-phase extraction method out-performs the classical analysis at any point.

The online concentration is therefore an ideal method for the concentration of analytical compounds, as it is a robust technique, quick, and it allows the determination of diminutive amounts of compounds in PAH oxidation mixtures.

The concentration and separation of a complex mixture is only the first step of any successful identification of organic compounds; the detection method is equally important. In the following, the application of UV spectroscopic detection is discussed.

24 Table 5. LOD and recoveries of PAH and OPAC with a 5-mL injection volume with online concentration, compared to traditional liquid-liquid extraction with GC-MS analysis

Micro Solid-Phase Liquid/Liquid Extraction Extraction GC-MS Analysis* HPLC-UV/DAD Analysis

LOD Recovery LOD Recovery [µg/l] [%] [µg/l] [%]

1-Indanone 0.21 87.2 420 66.2 1,4-Naphthoquinone 0.21 86.7 340 71.8 1-Hydroxynaphthalene 0.48 91.1 420 68.7 9,10-Phenanthrenequinone 0.33 90.1 700 58.6 6H-Dibenzo[b,d]pyran-6-ol 0.14 93.0 500 75.3 9H-Fluorene 0.11 90.7 180 73.9 Phenanthrene 0.07 90.7 420 70.3 * 5 mL sample extracted into 100 µL dichloromethane and 2 µL GC-MS injection volume

2.3. UV Absorption as Detection Method for Oxygenated Polycyclic Aromatic Compounds

UV absorption as a detection method is one of the work-horses in liquid chromatography.

It is simple, robust, and it allows for quantification over a relatively large concentration range

(typically 2-3 orders of magnitude). There are two basic UV absorption detectors available for chromatography, the single- or selectable- wavelength detector, which uses monochromatic light for irradiation, and the more advanced diode-array detector. The single- or selectable wavelength detector allows the recording of a chromatogram (signal intensity vs. elution volume or time in case of constant flow rate) only at a specified wavelength, whereas the diode-array detector enables the collection of UV spectral data for every point in the chromatogram. As most (known)

UV oxidation products of PAH retain the UV absorbing aromatic core, UV absorbance detection

25 is a very useful method for the detection and identification of polycyclic aromatic compounds.

Especially for the identification of PAH it is an extremely beneficial technique, as the UV absorbance spectrum of a PAH acts as its ―fingerprint,‖ enabling isomer-specific detection and identification.

For oxygenated compounds the situation is rather complicated. Although the oxidation usually enhances the UV absorption by adding an additional (oxygen-containing) chromophore to the system, the UV absorbance spectra of OPAC is commonly less distinct and less feature- full compared to the ones of PAH, hindering their identification. Figure 10 shows the normalized

UV spectral absorbance of pyrene and three pyrenequinones, namely the 1,6-, 4,5- and 1,8- isomers. Pyrene, the black curve in Figure 10, shows a typical PAH spectrum where distinctive, sharp features are visible. The oxygenated compounds either can exhibit similar features as can be seen in the case of 4,5-pyrenequinone (green spectrum), or their spectra is rather smooth with less sharp features as in the case of 1,6- and 1,8-pyrenequinone (red and blue spectra).

UV spectra of oxygenated compounds always depend on the solvent they are measured in, and their features can vary by a few nm. In the case of oxygenated compounds, in particular acids, acid anhydrides, acetals and ketals, an additional complication arises through the dependency of the configuration of the molecule on its hydration state and on the pH value of the solvent. An example case where this problem might arise is naphthalic acid anhydride (a photooxidation product of acenaphthylene), whose UV spectrum is shown in Figure 11 in red.

Figure 11 also shows the UV absorbance spectra of naphthalic acid (black) and the naphthalic acid dianion (blue). Naphthalic acid anhydride can hydrate to yield naphthalic acid, which— depending on the pH value of the mobile phase or the sample—can form anions. Therefore, the chromatographic separation could yield from one single compound, such as naphthalic acid

26 anhydride, up to three different peaks. In the current case, naphthalic acid anhydride can be identified by its somewhat distinct UV spectrum, but in the case of the acid or the anion, it would be challenging to correctly identify the compound from the UV spectrum or distinguish it from its hydration products.

Absorbance

200 300 400 500 600

Wavelength [nm]

Figure 10. UV absorbance spectra of pyrene (black), 4,5-pyrenequinone (green), 1,8- pyrenequinone (blue) and 1,6-pyrenequinone (red)

Despite all the potential difficulties, UV absorption as detection method is useful for the identification of OPAC. In the case of an OPAC with a distinct UV spectrum, such as 4,5- pyrenequinone, a spectral match of its spectrum with that of an reference compound will allow for the identification of that OPAC. Most OPAC exhibit rather feature-less UV absorbance

27 spectra, providing only limited information for their structural elucidation. Due to this lack of features and the wide similarities between different compounds’ UV spectra, any suspected identification will always require a confirmation with the chromatographic retention time of a reference compounds.

Absorbance

200 300 400 500 600

Wavelength [nm]

Figure 11. UV absorbance spectra of naphthalic acid anhydride (red), naphthalic acid (black) and naphthalic acid dianion (blue)

As such the need of a complementary detection method, which provides more detailed structural compound information, is evident. Mass spectral data would be the most desirable complementary method as it provides additional clues (molecular weight and structural

28 fragments) for the identification. In the following chapter the development of an advanced mass spectral technique coupled to HPLC for the identification of PAH and OPAC is discussed.

2.4. APPI-MS

Mass spectrometry is one of the most useful techniques for the structural elucidation and quantification of components in organic mixtures, especially when coupled to high-performance liquid chromatography (HPLC). A mass spectrometer allows the determination of the mass-to- charge ratio (m/z) of ions generated from the compound in question. Three different types of ionization techniques are generally used in HPLC instruments coupled to mass spectrometers.

Early established ionization techniques like electrospray ionization (ESI) and atmospheric- pressure chemical ionization (APCI) are effective in the analysis of moderately and strongly polar analytes97 but demonstrate limited effectiveness with nonpolar compounds. Atmospheric- pressure photoionization (APPI) was introduced in 2000 by Bruins et al.98 and Syage et al.99 as an ionization technique complementary to ESI and APCI. APPI is a soft ionization technique for both nonpolar and polar components, utilizing high energetic ultraviolet light (8.4-11.7 eV)100 to yield ions with only limited fragmentation. Several reviews101-104 regarding this new technique have demonstrated APPI as a ―promising‖ alternative to ESI and APCI, especially for nonpolar molecules such as polycyclic aromatic hydrocarbons, as well as their oxygenated counterparts.

Although APPI-MS has shown great promise as a ―universal‖ detection method, suitable for polar and nonpolar compounds, it is commonly deployed only as a quantitative tool. The qualitative utilization of APPI-MS has been so far limited to the molecular weight determination of PAH16,105-106 and has been neglected for other compounds. In this work it was desired to expand the application of APPI-MS towards the identification of unknown, polar compounds.

29 There are fundamentally different requirements applying to qualitative and quantitative mass spectrometry.

Background – The main requirement in case of quantitative mass spectrometry is that the background ions do not coincide with the monitored target ions. In qualitative mass spectrometry, the background can be critical for proper identifications and requires a low background, with few ions, especially in the expected m/z range of the unknown. The background should be consistent throughout the chromatogram and should depend little on the composition of the mobile phase. As the usual, non-volatile compounds analyzed with HPLC are relatively large molecules, a convenient background is below m/z 100.

Ionization efficiency – In quantitative mass spectrometry, high ionization efficiency for the analyte is desired to allow for low limits of detection. The compounds analyzed belong usually to one compound class, therefore the ionization can be optimized for a specific group of compounds e.g. PAH. In qualitative mass spectrometry, a high ionization efficiency is also desirable; however, the ion source must be able to ionize a wide range of different organic compounds, as the compound type of the analyte is unknown. In the case of OPAC, the type of the compound can vary widely from alcohols, ketones, aldehydes, carboxylic acids, esters, etc.

Mass spectrum – In quantitative HPLC-MS the observed ions do not necessarily need to correspond to the actual molecular configuration of the analyte. As long as the observed ions are indicative of the concentration and free of interferences, ions other than the molecular ion (e.g. metal ion adducts) can be utilized to quantify compounds.107 In qualitative analysis; however, a unique molecule-related ionization pattern, which relates directly to the elemental composition of the unknown compound, is required. Effects such as solvent adduct formation, chemical transformation, and double charging are undesirable, as they complicate the mass spectral

30 analysis. In quantitative work, a single ion from a compound is preferred as it allows for maximum sensitivity, whereas (moderate) fragmentation is a desirable property in qualitative mass spectrometry as it enhances the structural information available.

Negative and positive ionization - In quantitative mass spectrometry it is preferred if only one type of ion is formed, as it enhances sensitivity. For qualitative work the occurrence of both ion polarities can be very desirable, as it doubles the information gained.

Linearity is for quantitative utilization of mass spectrometry a necessity. For qualitative analysis, the linearity of the mass spectral signal is not the most important criterion, it is though still desirable. So that a consistent spectrum across the varying concentration of a chromatographic peak can be obtained.

Sensitivity is for both types of operation, quantitative and qualitative measurements, a significant parameter. For quantitative methods, high sensitivity translates directly into low limits of detection; for qualitative analysis, it can permit the analysis of minor products without the need for substantial amounts of sample.

2.5. APPI-MS Method Development

There are two APPI ionization sources commercially available, and they follow different design principles, as illustrated in Figure 12. The so-called ―linear‖ source (PhotoSpray®), depicted in Figure 12a, adheres to Bruins et al.’s design,98 in which the mass spectrometer is in line with the spray. The so-called ―orthogonal‖ source (PhotoMate®) devised by Syage et al.99 and depicted in Figure 12b, orients the capillary leading to the mass spectrometer in an angle that is 90 degrees to the spray. Although photoionization is common to both, the two sources differ significantly in their principal design: The linear source is intended to operate with an ionization- enhancing dopant, whereas the orthogonal source allows for direct ionization. Table 6 presents

31 an overview of the most commonly proposed ionization pathways for the formation of positive and negative ions in APPI. In reality, the ionization is a fairly complex mechanism, where several processes occur simultaneously.104,108-116 In the early years of APPI-MS there was somewhat of a dispute over which mechanism (direct or dopant), and ultimately which source, is the preferential one.117 Now (2011) it seems the opinion has swayed toward methods with dopant addition.118-119

The source utilized in this work is of the orthogonal design, namely the PhotoMate developed for the Agilent 1100 HPLC systems. The first step in any APPI-MS method development is the optimization of the ion source for the specific chromatographic conditions, the effluent type and its flow rate.

Table 6. Ionization mechanisms104,108-116 in APPI mass spectrometry

Positive Ion Formation Negative Ion Formation

Direct Ionization Electron Capture A + hυ → A●+ + e- A + e- → A-

Dopant Ionization Electron Generation D + hυ → D●+ + e- D + hυ → D●+ + e-

Charge Exchange Ionization by Oxygen ●+ ●+ - ●- D + A → A + D O2 + e → O2 ●- - ● Proton Transfer A + O2 → [A-H] + HO2 ●- ●- ●+ ● + A + O2 → A + O2 D + n S → [D-H] + SnH ●- - ● + ●+ S + O2 → [S-H] + HO2 SnH + A → [A+H] + n S - - D●+ + A → [A+H]●+ + [D-H]● A + [S-H] → [A-H] + S Analyte A, Dopant D, Solvent S

32 a UV Source Kr Lamp 10/10.6 eV Heater To Mass Nebulizer Gas Spectrometer HPLC Effluent

Dopant

Quartz Tube

b HPLC Effluent

Nebulizer Gas (N2) (+ Dopant) Nebulizer Needle and Heater Block

UV-Source Kr Lamp 10.0/10.6 eV

Drying Gas

To Mass Spectrometer

Vent Figure 12. Two commercially available ionization sources: (a) linear PhotoSpray® and (b) orthogonal PhotoMate®

2.5.1. APPI Source Optimization

The APPI source is the ion-generating interface connected to the mass spectrometer

(MS). As such, the source parameters, nebulizer pressure, vaporizer temperature, drying gas flow

33 rate and drying gas temperature, have to be optimized to yield an optimum amount of ions available for analysis. A convenient way of optimization is by means of flow-injection analysis

(FIA), which is the injection of a sample into a solvent stream without chromatographic separation. Naphthalene was chosen as a model compound and the FIA conducted in methanol as mobile phase at 0.2 mL/min flow rate, as most photooxidation products of PAH elute close to pure methanol conditions.

Figure 13 shows the dependency of the observed peak area of the m/z 128 ion

(naphthalene) on the nebulizer pressure. The peak area has been normalized to the maximum value measured. The nebulizer pressure was varied from 20 to 60 psig and showed a slight maximum at 35 psi. Higher nebulizing pressures will provide practical identical nebulization, but also use more nitrogen.

The next parameter to be optimized was the vaporizer temperature, which was tested from 250 to 500°C. Figure 14 shows the dependency of the normalized peak area of naphthalene on the vaporizer temperature. For this parameter a clear optimum of 350°C was observed.

The drying gas temperature and its flow were simultaneously tested from 200 to 350°C and 3 to 13 L/min respectively. Figure 15 depicts the naphthalene signal’s dependency on the drying gas flow, with each curve in Figure 15 representing a different drying gas temperature.

Independent of the drying gas temperature, a flow rate of 9 L/min provides optimal ionization.

Figure 15 reveals that a drying gas temperature of 350°C yields the best signal.

Following the analyses of Figures 13 to 15, the optimum spray settings of the APPI-MS source at a flow of 0.2 mL/min methanol were determined to be a nebulizer pressure of 35 psig, a vaporizer temperature of 350°C and a drying gas flow of 9 L/min with a drying gas temperature of 350°C.

34

1.0

0.8

0.6

0.4

Relative Signal Peak Area Peak Signal Relative 0.2

0.0 10 20 30 40 50 60 Nebulizer Pressure [psig]

Figure 13. Naphthalene (m/z 128) ionization dependency on the nebulizer pressure

The next step in the APPI-MS optimization is the optimization of the ionization parameters. Although in APPI-MS, the temperature plays a significant role in the ionization,120 the main parameters are nonetheless the capillary voltage and the fragmentor voltage. The capillary voltage is the voltage which drives the ions into the mass spectrometer (in addition to the vacuum), and it is applied from the spray chamber towards the mass spectrometer across the capillary. The fragmentor voltage affects the signal in two ways: First, it is responsible for fragmentation, capable of splitting ions into smaller fragments, and second, it facilitates the penetration of larger ions through the capillary. Figure 16 shows the effect of the capillary voltage on the analysis of two PAH and several OPAC. For each of the compounds the

35 dependency is slightly different, but nearly all compounds exhibit an optimum near 1200 V capillary voltage. Although the optimum capillary voltage for each of these compounds is within a close range (600 – 1600 V), it is easily conceivable that one particular setting might be less effective for one compound than for another. This effect would be beneficial for quantitative determination, as it allows to reduce the signal intensity of potential interfering ions. For qualitative work, it is a significant problem, as certain unknown compounds might be left undetected due to a different capillary-voltage optimum.

1.0

0.8

0.6

0.4

Relative Signal Peak Area Peak Signal Relative 0.2

0.0 200 300 400 500 Vaporizer Temperature [°C]

Figure 14. Naphthalene (m/z 128) ionization dependency on the vaporizer temperature

Another detrimental influence of the capillary voltage on the mass spectral signal can be seen in Figure 17. Figure 17 depicts the abundance of the individual ion signals of the mass spectrum of 9,10-phenanthrenequinone obtained at different capillary voltages. There are a total

36 of four ions that can be attributed to the presence of 9,10-phenanthrenequinone. The ion at m/z

209 is the hydrogen adduct ion [PHQ+H], which is a desired ion for qualitative work. The ion at m/z 181 originates from CO abstraction of the quinone and is also useful for structural elucidation. The ions at m/z 231 and 439 are adducts with sodium ions. These adduct-ions pose a severe problem for the utilization of APPI-MS for qualitative work, as they are the dominant main ions and they do not represent the molecular structure correctly. If 9,10- phenanthrenequinone (MW=208 Da) would have been an unknown compound in a sample, the main ion of m/z 231 would have suggested a molecular weight of 232 or 230 (for a system limited to carbon, hydrogen and oxygen), most likely preventing any chance of successful identification.

1.0 Drying Gas Temperature 350°C 300°C 0.8 250°C 200°C

0.6

0.4

Relative Signal Peak Area Peak Signal Relative 0.2

0.0 2 4 6 8 10 12 14 Drying Gas Flow Rate [L/min] Figure 15. Efficiency of naphthalene ionization as a function of drying gas flow and drying gas temperature.

37 The dependency of the ionization efficiency of compounds on the capillary voltage has been reported before,108 and the dependency does not affect the application of APPI-MS as a quantitative method.103,121 However, the presence of these effects would render fruitless the application of APPI-MS as a qualitative method.

600x103 6H-Dibenzo[b,d]pyran-6-one 9,10-Phenanthrenequinone 500x103 1-Indanone 9H-Fluorene Phenanthrene 400x103 Naphthalene 9H-Fluorene-9-one 1-Naphthol Indanole 300x103 1-Naphthoic acid Phenol

200x103

Signal / Peak Area Signal

100x103

0 0 500 1000 1500 2000 Capillary Voltage [V]

Figure 16. Peak area obtained from extracted positive ion chromatograms [M-2 to M+2] of polycyclic aromatic compounds versus the capillary voltage.

A potential solution for the successful application of APPI-MS as a quantitative method is the addition of an appropriate dopant. Hanold et al. have shown that by addition of an ionization-enhancing dopant, the dependency of the signal on the capillary voltage is greatly reduced.108 As an added benefit the signal intensity is also greatly enhanced, providing improved

38 sensitivity. As such, the next step in the method development was the design and the construction of a dopant delivery system, suitable for the current chromatographic method.

16x106

m/z 209 [PHQ+H]+ 6 14x10 + m/z 181 [PHQ-CO+H] m/z 231 [PHQ+Na]+ 6 12x10 m/z 439 [2 PHQ+Na]+

10x106

8x106

6x106

Signal / Peak Area Signal 4x106

2x106

0 0 200 400 600 800 1000 1200 1400 1600 1800 Capillary Voltage [V]

Figure 17. The effect of the capillary voltage on the ionization pattern of 9,10- phenanthrenequinone without a dopant.

2.5.2. Dopant Delivery System

Dopant addition in the orthogonal source has been reported in the literature106,108,118,122-126 to increase the sensitivity of the analysis significantly, by up to two orders of magnitude.108,123,126

An added benefit is a capillary-voltage-independent ionization.108 In these referenced cases, the dopant is delivered as a liquid, and the delivery itself is accomplished either by post-column addition,118,123-126 by using a suitable solvent as mobile phase,106 or by using a large (0.1 mL/min)

39 dopant flow into which a smaller, micro-HPLC effluent (50 μL/min) is mixed.122 The typical difficulties of these methods are the addition of dead volume to the chromatographic system, insufficient mixing, altering of the chromatographic separation, and/or the need for fairly complicated and excessive hardware. Active delivery systems like syringe pumps or HPLC pumps are necessary, and they lack the capacity for routine analysis or are fairly excessive equipment for this purpose. The chemicals most commonly used as the liquid dopants in these systems are benzene,106 toluene,108,126-127 propanone,108,122,126-127 anisole,126 and mixtures123,126 of these substances. From the miscibility properties of some of these compounds, other problems can arise: e.g., benzene and toluene are each incompatible with the highly aqueous mobile phase utilized in the chromatographic method developed in Section 2.1.

In order to avoid the chromatographic, hardware, and miscibility problems associated with liquid-dopant delivery, a new gas-phase dopant delivery system for use with HPLC-MS has been designed. As reported in the following, the newly designed passive dopant delivery system is based on the Syagen Technologies (Tustin, CA, USA) APPI source and Agilent (Wilmington,

DE, USA) mass spectrometers and provides great flexibility regarding the choice of mobile phase and dopant, without any of the previously mentioned disadvantages of liquid dopant systems. The basic performance of the dopant-assisted APPI-MS system is tested with polycyclic aromatic hydrocarbons (PAH) as well as OPAC.

2.5.2.1. Design of the Dopant Delivery System

The major technical challenge for HPLC-MS instruments lies in the coupling of the liquid effluent with a mass spectrometer that requires high vacuum. Practically all HPLC-MS instruments create a fluid spray with nitrogen and draw only a small stream of ions through a

40 capillary into the mass spectrometer. In the case of the Syagen Photomate® APPI interface, the nitrogen is delivered at high pressure (15-60 psig) and fairly large flow rate (1.9-2.1 L/min) in the nebulizer assembly, as depicted in Figure 12b. The nebulizer gas stream passes through 1/8‖

PTFE tubing that comes out of the mass spectrometer and leads back into the APPI source. By placing a series of stainless-steel bubblers in this accessible line, the nitrogen stream can be utilized as a carrier for dopant addition.

As depicted in Figure 18, a schematic view of the designed gas-phase dopant delivery system, the nitrogen flow can be either directed into the bubblers or the system can be set for dopant-free operation (for common maintenance procedures like source cleaning or tuning of the mass spectrometer). If the three-way valves are switched for operation with dopant, the nitrogen stream is split, to allow for the control of the dopant load. The majority of the flow is channeled through a flow restriction as a bypass stream. The other part of the flow (regulated through the metering valve) passes through the first heated stainless-steel bubbler, which preheats the gas.

The second bubbler is filled with the dopant material and acts as a saturation bubbler, where the nitrogen is loaded with the dopant. The third bubbler and the down-stream connecting tubing are set at a higher temperature than the saturation bubbler, preventing condensation upon mixing with the bypass stream. After mixing with the bypass stream, the nitrogen loaded with dopant flows into the APPI interface of the mass spectrometer.

The design of the components in Figure 18 is oriented towards the optimum range of mobile-phase flow rate for the orthogonal source, which has been reported to be between 0.2 and

0.4 mL/min.108,128 Our efforts have been focused on 0.2 mL/min mobile phase flow rate, ideal for the use with narrow-bore HPLC columns (2.1 mm diameter). Dopant is usually deployed at a flow rate of 10 to 20% of the mobile phase flow rate.114 An effluent flow rate of 0.2 mL/min

41 would therefore require 20-40 μL/min of dopant flow, or in the case of benzene, 17.5 to

35 mg/min. The usual operating pressure of the nebulizer gas stream is between 15 and 60 psi, delivering 1.9-2.1 L nitrogen per minute. At this flow rate, a full saturation of the nitrogen stream with benzene as dopant would amount to a dopant flow of 0.38 mL/min (nebulizer pressure,

35 psig; vapor pressure of benzene at 25°C, 1.84 psig129). A dopant flow of 0.38 mL/min can be tolerated by the source, but a large amount of dopant would be consumed, exceeding the mobile phase flow (0.2 mL/min). To avoid the high dopant consumption, only a part of the gas stream is loaded with the dopant in a bypass system, as shown in Figure 18. The control of the split between pure nitrogen and the dopant-saturated stream is accomplished by utilizing a low cv- value flow restriction (two 1/8‖ to 1/16‖ unions) and a metering valve, which has been sized accordingly (cv 0-0.03) to provide an approximate split ratio of 1:10.

Figure 18. Schematic of the gas-phase dopant delivery system

42 The main components of the dopant delivery system of Figure 18 consist of three pressurizable stainless-steel bubblers, which have been devised in-house and manufactured by the Department of Chemical Engineering machine shop. Each bubbler is made out of a simple three-inch diameter stainless-steel (SS 316) pipe, four inches long and welded at the bottom to a stainless-steel plate. The head of each bubbler is finished with a ―cam and groove‖ socket and dust cap, into which inlet, thermocouple well, and outlet have been welded. Instead of the standard rubber gasket, a fluorinated ethylene propylene (FEP) encased Viton® ring is used, to provide solvent and temperature resistance up to 200°C. This construction is superior to commercially available bubblers that utilize screw tops, as screw tops wear out much faster than the ―cam and groove‖ top. Connecting tubes are made of 1/8‖ stainless-steel tubing (SS 316).

As the vapor pressure and therefore the evaporation/delivery rate are strongly temperature dependent, two heating tapes are wrapped around the bubblers and connecting tubing, in order to ensure consistent temperature and to enable the vaporization of less volatile dopants. The first heating tape covers the first bubbler, where the gas is preheated, and the second bubbler, the saturator. The second heating tape, wrapped around the third bubbler and the connecting tubing, ensures that the mixture is heated beyond its dew point—especially at the point of mixing with the bypass nitrogen, where condensation could otherwise occur if the saturation temperature were significantly higher than ambient. Both heating tapes are controlled by PID temperature controllers—connected to type K thermocouples, positioned as indicated in

Figure 18—which allow the two temperatures T1 and T2 to be controlled to within ±1°C.

The saturation bubbler, bubbler #2 in Figure 18, containing the dopant, features a shortened dip tube, so that the nitrogen, instead of bubbling through the solvent like a ―real‖ bubbler, only blows onto it. This feature is necessary to avoid oscillating delivery of dopant into

43 the APPI-interface, which occurs with bubble formation. Naturally, as the dopant is continuously used, the level of liquid falls and the blowing becomes less efficient in terms of its mass transfer, resulting in changing delivery levels. A constant flow rate can be achieved, nevertheless, through the use of a PTFE tube, which along its length is sliced half-open, to guide the nitrogen stream onto the surface.

The setup shown in Figure 18 is designed for highly volatile dopants like those most commonly used—e.g., benzene, toluene, anisole, and acetone (propanone). If substances with lower volatility are to be used, the system can be fairly easily reconfigured by replacing the flow restriction with a low-cv-value metering valve and eliminating the metering valve on the saturation stream side. Alternatively, a higher temperature can be used to ensure sufficient vaporization.

The entire system is rated to 90 psi and has been tested at temperatures up to 120°C. The saturator bubbler has a total capacity of 450 mL, which provides for continuous operation for more than two weeks (0.2 mL/min effluent, 20 μL/min dopant flow rate) before the system has to be recharged.

The basic delivery performance of the dopant delivery system in Figure 18 was tested with benzene as dopant at 40°C saturator temperature (T1 in Figure 18) and 50°C for the down- stream flow path (T2). (The latter temperature was chosen to eliminate the influence of the local thermal disturbance from the heat vent of the mass spectrometer.) The amount of dopant delivery was assessed by loading the bubbler #2 with a 100-mL glass bottle filled with benzene and measuring the loss several times after 100 to 500 min. To account for the bias during opening and pressurizing the system, the consumed mass was plotted against time, and the slope was taken to be the delivery rate of benzene. Figure 19 shows the dependence of the dopant delivery

44 rate on the metering valve setting. Figure 19 illustrates that, when the metering valve is open for

≥4 turns, the measured benzene flow rate varies regularly with the number of turns. The deviations that occur at ≤4 turns are inconsequential since they correspond to benzene flow rates that are lower than what is required for dopant delivery. We therefore see that the passive gas- phase dopant delivery setup is capable of delivering benzene with a mass flow rate of up to 40 mg/min with no more than 10% deviation.

2.5.2.2. Dopant Selection

The correct choice of a dopant for qualitative mass spectrometry is an important issue, to which the in Section 2.4. listed principles of qualitative mass spectrometry apply. The most important criterion is the need to provide effective and consistent ionization for a large group of different compounds. For this purpose several PAH and OPAC were analyzed with dopant- assisted HPLC-APPI/MS. The ionization efficiency of the main ions of these compounds were tested with the most commonly utilized dopants: benzene, toluene, acetone and anisole. Figure

20 shows the differences in ionization efficiency when using these dopants for phenolic compounds such as phenol and 1-naphthol, ketones such as 9,10-phenanthrenequinone, ester such as 6H-dibenzo[b,d]pyran-6-one and PAH such as 9H-fluorene and phenanthrene. The graph shows the peak area obtained from the extracted-ion chromatogram for each compound’s molecular ion (±2 Da, to account for protonation and deprotonation of the main ion) in the positive and negative ionization modes—with the exception of 1-naphthoic acid, whose main negative ion at m/z 127 was used for the negative ionization efficiency.

A comparison of Figure 20a and Figure 20b reveals, that all signals are greatly enhanced by the presence of a dopant. This observation has to be treated with caution as the ionization

45 parameters utilized were not optimized for the specific compounds. The utilization of optimized ionization parameters for dopant-free operation could have potentially yielded higher signals.

50

40

30

20

Benzene Flow Rate (mg /min) Rate Benzene Flow 10

0 1 2 3 4 5 6 7 8 9

Number of Turns Open of the Metering Valve

Figure 19. Benzene delivery rate at 40°C saturator temperature.

Figure 20b shows, that not all tested compounds produce a negative ion, and there is little effect of the dopant on the ionization efficiency of negative ions. This observation is consistent with the suggested ionization mechanism of these compounds via electron absorption (shown in

Table 5). As all of the tested dopants produce electrons in abundance, little difference is seen between the individual dopants.

46 The contrary is the case with respect to the positive ionization efficiency. Acetone and anisole are very efficient dopants for some compounds (1-hydroxynaphthalene, 9,10- phenanthrenequinone and 6H-dibenzo[b,d]pyran-6-one); however, they lack the universality toluene and benzene have to offer. Benzene was the only tested dopant that enhanced the ionization of all compounds, indicating its universal applicability for unknown compounds.

Therefore for qualitative determinations, benzene appears to be a good choice as dopant, as it allows for the widest range of compounds to be analyzed.

For qualitative mass spectrometry another criterion arises from the background ions dopants generate. These ions, which could mask an analyte’s signal, are required to be low in m/z to avoid potential interference with signals from the ions of the compounds of interest. The background also ought to be unaffected by changing mobile-phase compositions, to allow for gradient elution. It should also be low in intensity to allow for easy detection of the analyte’s signal.

Of the tested dopants, acetone and anisole provide the most consistent backgrounds. With water and water/methanol as mobile phase, acetone produces a single ion at m/z 59, which is very low. The background mass spectrum of acetone is dependent on the mobile phase composition, adding other low-m/z ions at 73, 87 and 92 with increasing methanol content. In the negative ionization mode acetone produces no measurable background.

Anisole is the compound with the most constant mass spectrum with respect to its background ions; however, its main ions at m/z 108, 78, 65 and 124 are at significantly higher values of m/z than are the ones of acetone. There are also some background ions observed in the negative ionization mode (though consistent) at m/z 110, 108, 93 and 60. A disadvantage common to both acetone and anisole is the high level of the ionization of the dopant. Anisole and

47 acetone produce a background signal with a total-ion current that is 4-8 times as intense as the one generated by benzene or toluene.

Anisole 106 (a) Positive Ionization Acetone + Benzene Toluene

-[M+2]

+ No Dopant 105

104

Peak Area EIC [M-2]

103 105 (b) Negative Ionization

-

-[M+2]

- 104

103

Peak Area EIC [M-2]

102

Figure 20. Positive (a) and negative (b) ionization efficiencies with anisole, acetone, benzene and toluene as dopants; for the negative ions of 1-naphthoic acid m/z 125-129 was used instead of m/z 170-174

48

Toluene and benzene both exhibit strongly solvent-dependent backgrounds. In a pure aqueous mobile phase, both possess strong background signals, whose intensity reduces significantly with increasing methanol concentration. Toluene shows positive ions at m/z 108 and

56 in a water/methanol mobile phase, whereas additional ions at m/z 73 and 87 appear in pure methanol (with 87 being the most abundant ion). The negative ion signal is only minimally affected by ions at m/z 108 and m/z 119.

Benzene’s ionization also changes significantly with the mobile phase composition— from a single ion at m/z 94 at highly aqueous conditions, to m/z 78, 87 and 56 at mixed solvent conditions, to predominantly the ion of m/z 87 (minor ions at m/z 78, 74, 55) in pure methanol.

None of the tested dopants are perfect with regard to their utilization in qualitative mass spectroscopy. The good pattern stability of the ions of anisole and acetone is offset by their high ionization levels, whereas the low m/z ratios and lower intensities of benzene and toluene are offset by these aromatics’ strongly changing backgrounds.

For determining polycyclic aromatic compounds, benzene is the dopant of choice, as it allows the ionization of the widest range of compounds. Although its background signal changes with mobile phase composition, its low m/z ratio does not intrude into the typical m/z range of polycyclic aromatic compounds (>m/z 128).

2.5.2.3. Ionization Performance of Benzene as Dopant

Benzene was tested as dopant in conjunction with four PAH (naphthalene, 9H-fluorene, phenanthrene, anthracene) as the APPI test substances. By adding 40 mg/min benzene as dopant to the APPI-MS, a 20–fold increase in the sensitivity for the analysis of PAH was achieved.130

Apart from this significant sensitivity gain for PAH analysis, the benzene addition is also capable

49 of eliminating the capillary-voltage dependency of the ion signal of oxygenated compounds.

Figure 21 shows the signal intensities of the same PAH and OPAC as they were tested without dopant in Figure 16. As can be seen by comparing Figure 21 with Figure 16, the addition of benzene as dopant causes the sensitivity for all compounds to be increased by at least an order of magnitude. Furthermore the capillary voltage does not yield an apex for the signal intensity any more, as all signals flatten out with increasing capillary voltage. This is a very important improvement as the enhanced sensitivity and the compound-independent ionization are essential for the qualitative assessment of OPAC via APPI-MS.

30x106 6H-Dibenzo[b,d]pyran-6-one 9,10-Phenanthrenequinone 1-Indanone 25x106 9H-Fluorene Phenanthrene Naphthalene 9H-Fluorene-9-one 6 20x10 1-Naphthol Indanole 1-Naphthoic acid 15x106 Phenol

Signal / Peak Area Signal 10x106

5x106

0 0 500 1000 1500 2000 Capillary Voltage [V]

Figure 21. Peak area obtained from extracted positive ion chromatogram [M-2 to M+2] of polycyclic aromatic compounds versus the capillary voltage. Benzene was deployed as a dopant at 40 mg/min.

50 The previously mentioned (Section 2.5.1.) disadvantage of adduct formation of 9,10- phenanthrenequinone with sodium ions and its dependence on the capillary voltage are addressed in Figure 22. Figure 22 demonstrates that the ionization pattern of 9,10-phenanthrenequinone depends much less on the capillary voltage in the presence of benzene as dopant than without dopant. The adduct formation with sodium ions (ions 231 and 439) is greatly suppressed, and at any capillary voltage above 1400 V no adduct ion can be observed, allowing for the correct determination of 9,10-phenanthrenequinone’s molecular weight as 208 Da from the [M+H] ion at m/z 209.

6 25x10 m/z 209 [PHQ+H]+ m/z 181 [PHQ-CO+H]+ m/z 231 [PHQ+Na]+ 20x106 m/z 439 [2 PHQ+Na]+

15x106

10x106

Signal / Peak Area Signal

5x106

0 0 200 400 600 800 1000 1200 1400 1600 1800 Capillary Voltage [V]

Figure 22. The effect of the capillary voltage on the ionization pattern of 9,10- phenanthrenequinone [PHQ] with 40 mg/min benzene as dopant.

51 From the above results, it is clear that the addition of benzene as dopant is indeed the silver bullet for qualitative utilization of APPI-MS, as it allows for the capillary-voltage independent-ionization of a wide range of different compounds, exhibits a low m/z background, and prevents the formation of misleading sodium adduct ions. Without the addition of benzene as dopant, identifications of oxygenated species would be complicated at the very least, if not completely impossible.

The last ionization parameter that has yet to be considered for the optimization of the

APPI-MS is the fragmentor voltage. With dopant-assisted APPI-MS, there is only a limited dependency of the ionization efficiency of the tested compounds on the capillary voltage or the fragmentor voltage. Fragmentor voltages from 75 V to 200 V were tested. An increase in the fragmentor voltage increased the intensity of all ions, the ions of the target compounds as well as the ions associated with the background. Therefore, higher fragmentor voltage facilitates any ion’s migration toward the mass spectrometer. A fragmentor voltage of 150 V was selected as preferred value due to the best signal-to-noise ratio.

From the analyses of Sections 2.5.1 and 2.5.2, an optimized, final set of parameters for the APPI-MS sources can now be defined, concluding the APPI-MS method development for qualitative purposes. Table 7 shows the values of these finalized parameters for dopant-assisted

APPI-MS analysis of PAH and OPAC with methanol/water mobile phases at a flow rate of 0.2 mL/min.

With the enhanced and optimized method, dopant-assisted APPI-MS is now capable of detecting ions of oxygenated polycyclic aromatic compounds regardless of their compound class.

The following section discusses the observed ionization patterns and their potential use for qualitative analysis of OPAC.

52 Table 7. Optimized APPI-MS parameters for qualitative analysis at 0.2 mL/min mobile phase flow (methanol)

Parameter Value

Nebulizer pressure 35 psig

Vaporizer temperature 350°C

Drying gas flow rate 9 L/min

Drying gas temperature 350°C

Capillary voltage -1800 V / +1800 V

Fragmentor voltage 150 V

Dopant Benzene; 40 mg/min

2.6. APPI-MS Spectra

Structural elucidation of organic compounds based on mass spectrometry involves two main activities, the determination of the molecular weight and elemental composition of the compound from the molecular ion, and the deduction of the molecular structure from fragment ions. In particular GC-MS with electron impact ionization (EI) has been extensively used for this purpose, and large libraries such as the NIST library131 are readily available to aide in the identification of compounds. APPI-MS, in contrast to EI-MS, is considered a ―soft‖ ionization technique, yielding very few fragments and having been used predominantly for the determination of the molecular weights and the elemental compositions of compounds.105-106,132-

133

The elemental composition, derived from APPI-MS data, can be already very helpful by itself; however, it does not compare with the information EI mass spectrometry provides.

Nonetheless, the feature of APPI-MS to record negative and positive ions simultaneously enables

53 the user to collect additional information beyond the molecular weight, especially when fragmentation is observed. In the following, the observed patterns for several polycyclic compound classes are presented and discussed. The observed regularities of the mass spectra were tested with various representatives of each class and are demonstrated on select examples in the next section.

2.6.1. Polycyclic Aromatic Hydrocarbons

Polycyclic aromatic hydrocarbons, as nonpolar molecules, are the compound class for which APPI-MS was primarily developed. PAH are not easily amenable to ESI134-135 and are also difficult to assess with APCI.136-137 The literature106,118-119,126 documents several examples of the application of APPI-MS for both the qualitative and quantitative analysis of PAH.

Figure 23 shows the APPI-MS spectrum of 9H-fluorene (Figure 23a) and phenanthrene

(Figure 23b). Generally, PAH produce only positive ions. The red signal shows the positive ion trace, and the blue signal designates the negative ion trace. In both cases, 9H-fluorene and phenanthrene, the positive ions consist solely of the molecular ions and their hydrogen adducts.

The distribution between the [M] and [M+H] ions depends on several factors—including, but not limited to, temperature, mobile phase composition and the structure of the PAH itself.120

Phenanthrene does not yield any negative ions, whereas 9H-fluorene yields a very weak negative ion at m/z 165 (abundance <0.3%) . A negative ion is atypical for PAH and can be attributed to the high acidity of the hydrogen at the 9-position in 9H-fluorene, which is unique for 9H- fluorene and its benzologues.138

54 100 100 a 166 [M]+ b 179 [M+H]+ 80 80 60 60 40 40 20 20 165 [M]- 0 0 Relative Ion Abundance [%] Ion Abundance Relative 100 120 140 160 180 200 [%] Ion Abundance Relative 140 160 180 200 220 m/z m/z

Figure 23. APPI-MS spectra of 9H-fluorene (a) and phenanthrene (b) with benzene as dopant; the red line marks the abundance of the positive and the blue line the abundance of the negative ions

Fragmentation of these two PAH has not been observed and would not be expected for simple, unsubstituted PAH. Thermal re-arrangement or ring closure, especially when it is thermodynamically favorable (as is in the case of toluene to tropylium139 or ortho-terphenyls to triphenylenes), can occur. From these observations, it is possible to draw the conclusion that for

PAH only positive ions are expected, either the positive molecular ion [M] or its adduct [M+H].

2.6.2. Hydroxyl-Substituted Polycyclic Aromatic Hydrocarbons

Hydroxyl-substituted polycyclic aromatic hydrocarbons are a special class of alcohols, as the hydroxyl group is bound directly to an aromatic structure. Figure 24 shows the mass spectra of phenol (Figure 24a) and 1-naphthol (Figure 24b). Both compounds produce a molecular ion

[M] in the positive mass spectrum and exhibit a less intense [M-H] ion in the negative trace that signifies the abstraction of one hydrogen. This behavior is consistent with the known acidity of the hydrogen atom of the phenolic hydroxyl group,140 which allows the abstraction of the hydrogen atom. It is interesting that there is no significant occurrence of the positive [M+H] ion

55 for either compound in Figure 24, as each shows a clear preference for the [M] ion. Therefore hydroxyl-substituted aromatic compounds are expected to yield negative and positive ions, with

[M] and [M-H] as dominant ions in the positive and negative mass spectra respectively.

100 100 a 94 [M]+ b 144 [M]+ 80 80 60 60 40 40 - - 20 93 [M-H] 20 143 [M-H] 0 0 Relative Ion Abundance [%] Ion Abundance Relative 60 80 100 120 140 [%] Ion Abundance Relative 100 120 140 160 180 200 m/z m/z

Figure 24. APPI-MS spectra of phenol (a) and naphthalene-1-ol (b) with benzene as dopant; the red line marks the abundance of the positive and the blue line the abundance of the negative ions

2.6.3. Alcohols

Hydroxyl groups that are not directly bound to an aromatic body can exhibit significantly different behavior than their phenolic analogues. Figure 25 shows the mass spectra of 1-indanol

(Figure 25a) and 9-hydroxy-9H-fluorene (Figure 25b). As can be seen in Figure 25a, 1-indanol does not yield any negative ions and shows two main positive ions, [M] at m/z 133 and [M-H] at m/z 134. Additionally, there is a positive fragment ion observable at m/z 117, which corresponds to an [M-OH] ion. Figure 25b reveals that 9H-fluorene-9-ol shows an identical pattern of positive ions. Therefore for non-phenolic alcohols three positive ions are observed, the molecular ion at [M], the ―oxidized‖ ion at [M-H] and the fragment at [M-OH].

56 100 + 100 + a 133 [M-H] b 165 [M-OH] + + 80 134 [M] 80 181 [M-H] + 60 + 60 182 [M] 117 [M-OH] 40 40 20 20 0 0 Relative Ion Abundance [%] Ion Abundance Relative 80 100 120 140 160 [%] Ion Abundance Relative 140 160 180 200 220 m/z m/z

Figure 25. APPI-MS spectra of 1-indanol (a) and 9-hydroxy-9H-fluorene (b) with benzene as dopant; the red line marks the abundance of the positive and the blue line the abundance of the negative ions

2.6.4. Aromatic Ketones

Aromatic ketones can appear in two forms, either as quinoid structures, a benzenoid PAH with an even number of carbonyl groups, or as simple ketones (such as 9H-fluorene-9-one), with only one carbonyl group. Figure 26a shows the mass spectrum of 9H-fluorene-9-one and Figure

26b the mass spectrum of 9,10-phenanthrenequinone. As is evident in Figure 26a, 9H-fluorene-

9-one yields upon ionization only positive ions, with [M] and [M+H] as molecular ions present and a minor amount of a fragment at m/z 153 [M-CO+H], resulting from CO abstraction.

Figure 26b shows that 9,10-phenanthrenequinone yields positive and negative ions. The negative ion is a weak molecular ion at m/z 208 [M]. In the positive mass spectrum, the main ion is the protonated [M+H] ion. Additionally there are two minor positive ions at m/z 181 [M-

CO+H] and at 153 [M-2 CO+H] (not labeled in Figure 26b), which result from CO abstraction of the molecule. The CO abstraction is temperature-dependent and capillary-voltage-dependent (see

Figure 22) and, due to its low abundance, might not always be observed.

57 Therefore as a rule for the identification of quinoid structures and ketones, a [M] and/or

[M+H] ion and fragments due to CO abstraction are expected in the positive mass spectrum. For the negative mass spectrum of quinoid structures and ketones, a negative [M] ion is expected.

The mass spectrum of 9H-fluorene-9-one, which lacks negative ions, is an exception to this pattern, as other ketones, such as 1H-phenalene-1-one or 7H-benz[de]anthracene-7-one yield a negative [M] ion.

100 + 100 + a 180 [M] b 209 [M+H] 80 + 80 181 [M-H] 60 60 40 40 + + - 20 152 [M-CO] 20 181 [M-CO+H] 208 [M]

Relative Ion Relative Abundance 0 0 140 160 180 200 220 [%] Ion Abundance Relative 160 180 200 220 240 m/z m/z

Figure 26. APPI-MS spectra of 9H-fluorene-9-one (a) and 9,10-phenanthrenequinone (b) with benzene as dopant; the red line marks the abundance of the positive and the blue line the abundance of the negative ions

2.6.5. Carboxylic Acids and Esters

Carboxylic acids are commonly found oxidation products of polycyclic aromatic compounds, incorporating a functional group with the second highest oxidation state for carbon.

Figure 27a shows the spectrum of 1-naphthoic acid. The main positive ion is the molecular ion

[M] at m/z 172. Additionally, a positive [M-OH] fragment at m/z 155 is found, which is formed through OH abstraction. The carboxylic acid exhibits an intense negative mass spectrum, with the main ion being a fragment at m/z 127 ([M-COOH]), resulting from the expulsion of the entire

58 carboxylic group. Additionally a weak molecular ion [M-H] at m/z 171, formed through hydrogen abstraction can be observed.

A representative of molecules that are closely related to carboxylic acids, but with significantly different mass spectra, is 6H-dibenzo[b,d]pyran-6-one, a lactone, a cyclic ester.

Figure 27b displays the APPI-MS mass spectrum of 6H-dibenzo[b,d]pyran-6-one. The ester lacks any negative ions and yields, in the positive spectrum, only the molecular ion [M+H] as main ion.

For carboxylic acids, a dominant positive ion at [M] and minor ions at [M-OH] are expected, whereas for esters predominantly a protonated positive molecular ion [M+H] is expected. Carboxylic acids yield additionally a strong negative ion signal with the dominant ion at [M-COOH].

100 100 + a 172 [M]+ b 197 [M+H] - 80 127 [M-COOH] 80 + 60 155 [M-OH] 60 40 40 20 20 - 171 [M-H] 0 0 Relative Ion Abundance [%] Ion Abundance Relative 120 140 160 180 200 [%] Ion Abundance Relative 160 180 200 220 240 m/z m/z

Figure 27. APPI-MS spectra of 1-naphthoic acid (a) and 6H-dibenzo[b,d]pyran-6-one (b) with benzene as dopant; the red line marks the abundance of the positive and the blue line the abundance of the negative ions

59 2.6.6. Aldehydes

Aldehydes are very reactive molecules, and this fact is mirrored in their mass spectra.

Figure 28 shows the mass spectrum of trans-cinnamaldehyde (Figure 28a) and 1-napthaldehyde

(Figure 28b). Aldehydes exhibit significant fragmentation in the positive ion mass spectra and do not yield negative ions. Cinnamaldehyde and naphthaldehyde each have, as main ion, the positive molecular ion [M+H] and one fragment [M-CO+H], stemming from CO abstraction.

Cinnamaldehyde yields an additional ion via OH abstraction at m/z 115. 1-naphthaldehyde also yields another ion of m/z 171, which could originate either from oxidation with molecular oxygen or from the disproportion of the aldehyde to (1-naphtyl)methanol and 1-naphthoic acid

(Cannizzaro reaction). (The latter seems less likely, as any acid would generate ions in the negative mass spectrum; however, as the residence time in the APPI source is very short, the compound might not remain long enough to experience full ionization toward the negative ions.)

100 + 100 + a 133 [M+H] b 157 [M+H] + 80 80 129 [M-CO+H] 171 60 60 + 115 [M-OH] 40 40 + 105 [M-CO+H] 20 20 0 0 Relative Ion Abundance [%] Ion Abundance Relative 80 100 120 140 160 [%] Ion Abundance Relative 120 140 160 180 200 m/z m/z

Figure 28. APPI-MS spectra of trans-cinnamaldehyde (a) and 1-naphthaldehyde (b) with benzene as dopant; the red line marks the abundance of the positive and the blue line the abundance of the negative ions

60 For aldehydes, at least two positive ions are expected, the protonated molecular ion

[M+H], and ions resulting from abstraction of hydroxyl [M-OH] or CO [M-CO+H].

Additionally, another ion [M+O], stemming from oxidation or a disproportion reaction, can occur.

2.6.7. APPI-MS Spectra for the Identification of OPAC

The combination of all the observed ionization patterns allows the establishment of a set of rules, which aide in the identification of polycyclic aromatic compounds. Table 8 lists all, in

Section 2.6.1 to Section 2.6.6, derived rules for the ionization pattern of polycyclic aromatic compounds. These rules were established based on the observation of the ionization pattern of several representatives of each class, with illustrative examples given in Section 2.6.1 to 2.6.6.

For each compound class a unique pattern is observed. Therefore, dopant-assisted APPI-MS can be used to determine the types of functional groups a compound incorporates. Caution has to be applied when a molecule is tested with these rules. As they are only rules, there will be exceptions, as was demonstrated in the case of 9H-fluorene and 9H-fluorene-9-one. Therefore the APPI-MS source should not be considered as just an ionization source, but rather an ion- generating photochemical reaction chamber. Although the ion formation can be quite complex, for many structures the chemistry is quite predictable and will allow for a successful identification of the functional groups associated with the molecule.

The information obtained from the mass spectrum is useful in two ways, first in identifying the functionality of the molecule and second, in obtaining the molecular weight of the compound in question. The molecular weight is essential for the determination of the elemental composition. Even with the use of a low-resolution mass spectrometer, such as the single-

61 quadrupole mass spectrometer used in this work, it is possible to derive the elemental composition. As in this work any reactive compound incorporates only carbon, oxygen and hydrogen, only even-number molecular weights are possible. The exact composition can be further narrowed down by utilizing a program, such as the NIST isotope calculator,141 to determine all possible elemental permutations and choose the most likely composition based on the hydrogen deficiency.

Table 8. Observed APPI-MS ionization patterns of polycyclic aromatic compounds

Compound class Positive ions Negative ions Other characteristics

PAH [M], [M+1] --- Thermal rearrangement and condensation possible

Phenolic alcohols [M] [M-1] Positive more intense than negative signal

Non-phenolic [M], [M-1], [M-17] --- alcohols

Ketones [M+1], ([M-27]) ([M]) Negative signal is weak and can be absent, CO abstraction products is weak

Aldehydes [M+1], [M-27], --- Oxidation product [M+16] is not ([M+16]) always present, CO abstraction product is a strong signal

Carboxylic acids [M], [M-17] [M-45], [M-1] Negative ion is a strong signal and fragment [M-45] is dominant

Carboxylic ester [M+1], ([M]) ---

By combining the molecular formula with the ionization pattern’s ability to provide insight into the chemical functionality, the numbers of possible candidate structures for identification are greatly limited and enable a targeted synthesis or comparison with a reference compound. This provides an invaluable advantage for the identification of polycyclic aromatic

62 compounds, which is challenging due to the tremendous number of different isomers. Especially for the identification of products of PAH photooxidation, it is crucial to narrow the pool of potential candidates as much as possible, as otherwise identification attempts would require laborious synthetic efforts.

The application of the, in this chapter developed, analytical techniques toward the identification of OPAC formed during the photooxidation of PAH is described in Chapter 4. The description of the reactors used to obtain these photooxidation samples is outlined in the next

Chapter.

63 3. EXPERIMENTAL SETUP 3.1. Thin-Film Reactor

For the study of the interfacial uptake of PAH and their photochemical conversion, several types of reactors have been reported in the literature. Falling droplet reactors,142 fog chambers143-144 and thin-film reactors38,71-73 have been used to simulate the high surface-to- volume ratio of hydrometeors such as fog. In this work, a previously constructed thin-film reactor,80 designed for uptake experiments, was adapted to allow for uptake and photooxidation studies of PAH in thin water films from ambient to sub-freezing temperatures.

Figure 29 shows a schematic depiction of the experimental setup. Air (ultra-high-purity grade, Air Liquide America Specialty Gases LLC, Baton Rouge, LA, USA) is supplied from a compressed gas cylinder. The gas flow rate (70 mL/min) is controlled by a mass flow controller

(R60219, Omega Engineering, Stamford, CT, USA). The air stream passes through a packed-bed saturator (12.5 mm×250 mm stainless-steel tube), which is filled with acid-washed alumina beads (60-80 mesh, Chromosorb PAW, Sigma Aldrich, St. Louis, MO, USA). The alumina was charged with the desired PAH by soaking the beads in a saturated pentane solution of the PAH and letting the solvent evaporate to achieve a PAH loading of 10%. The saturator is heated to

35°C with a column heater (TCM 2000, Chromtech, Apple Valley, MN, USA) to provide a constant PAH-saturated gas stream independent of room temperature. The PAH-saturated stream passes first through a stainless-steel loop, which is immersed in the first chiller, chiller #1

(Polystat, Cole-Parmer, Vernon Hills, IL, USA) in Figure 29, reducing the concentration of the

PAH in the gas phase to the saturation point of the bath temperature. The saturated stream travels next into a custom-made bubbler with backflow (Widgett Scientific, Baton Rouge, LA, USA),

64 situated in the bath of chiller #1. The bubbler consists of two chambers, the first of which is filled with water and a small amount of the desired PAH. The gas stream is pushed through a porous frit in the first chamber of the bubbler to adjust the relative humidity of the incoming air stream to 100%, preventing drying-out of the tested thin water films. The second chamber impinges the humid air stream coming from the first chamber with the water surface to deposit any droplets which might have been carried over from bursting bubbles in the first chamber.

Both chambers are connected with each other at their bottom to allow the deposited water droplets to flow back. The water-and-PAH-saturated stream leads next into the thin-film reactor.

As the connection from the thermostatted bubbler to the reactor is relatively long (60 cm), the entire line is arranged as a shell-and-tube heat exchanger, depicted as heat exchanger #1 in

Figure 29. This setup ensures temperature consistency from the point of saturation to the point of entry into the thin-film reactor.

Figure 30 shows the cross-sectional view of the thin-film reactor. The thin-film reactor is a temperature-controlled boat reactor, built out of a 5 cm×101 cm Kimble KG-33 glass tube, which is covered on the inside with a PFA film (0.508 mm thickness, McMaster-Carr,

Robbinsville, NJ, USA) to prevent PAH adsorption onto the glass. A heat exchanger made out of copper, which is also covered with PFA foil, provides support and accurate temperature control of the glass boat. The heat exchanger is supplied with cooling fluid from chiller #1 in Figure 29.

The temperature of the glass slide is measured at the top of the heat exchanger with two type-T thermocouples (Omega Engineering, Stamford, CT, USA). The gas-phase temperature is also measured by two type-T thermocouples inside the reactor.

65 Chiller #1 Carrier gas: Air Stainless steel loop

Mass flow Heated PAH controller saturator Column

Bubbler with automatic back flow

Insulated enclosure Vent Chiller #2

TI TI VC Cooling air TI TI

Vent Heat exchanger #1

Temperature controlled boat reactor Gas phase sample collection VC = Vortex cooler (optional)

Figure 29. Thin-film reactor (schematic). Solid arrows represent gas flow; dashed lines represent coolant flow

2 x 2 UVB Lamps (ea 15 W) with external cooling

KG-33 Glass Tube, PFA-coated Etched Glass Boat Area: 270 cm2

Heat Exchanger, Copper, PFA-Coated

Figure 30. Cross-sectional view of the temperature-controlled boat reactor

66 The glass boat is a 40 mm×1000 mm×2-mm glass slide (Technical Glass Products Inc.,

Ohio, USA), into which a rectangular trough of 274 µm depth and 270.7 cm2 area has been etched with hydrofluoric acid. The slide has been subjected to surface treatment with dichlorodimethylsilane and sodium hydroxide to create a hydrophobic outside and a wettable trough, so that the water film is well contained. A detailed description of the etching of the slide and its surface modification can be found in Appendix B.

The reactor tube with the heat exchanger is housed within an insulated aluminum box, which acts as thermal containment. The temperature in this insulated enclosure is controlled by supplying dry, pressurized air, which has been cooled by a second chiller, chiller #2 (Polystat,

Cole-Parmer, Vernon Hills, IL, USA) in Figure 29, to the reactor temperature. The enclosure temperature is controlled in such a manner, that the deviation from the reactor temperature is only positive and no more than 1°C, to avoid condensation and transport of water from the slide to the reactor wall. For experiments conducted below 0°C, the cooling power of chiller #2 alone is insufficient to cool the enclosure; therefore a vortex cooler (Vortex Tube Mini, Arizona

Vortex Tube Manufacturing Co., Wickenburg, AZ, USA) is used to reduce the cooling-air temperature even further. This setup, designed for uptake experiments, allows for temperature control down to -9°C.

A typical uptake experiment consists of loading the reactor with the glass slide, filled with pure water (Burdick and Jackson LC-MS grade, Honeywell, Muskegan, MI, USA).

Depending on the amount of water, a different film thickness is achieved. One mL of water is equivalent to a film thickness of 36.9 µm. The PAH is introduced via the gas phase for at least five hours to allow for saturation of the reactor and the liquid phase. The uptake of the PAH into the water film is then measured by quickly opening the reactor and washing the water solution on

67 the surface of the slide with acetonitrile into a volumetric flask. The concentration of this solution is then measured via HPLC. Gas-phase concentrations are obtained by absorbing the gas stream for a set time period into ice-cooled acetonitrile and measuring the concentration via

HPLC.

For photooxidation experiments, UV light is delivered via two fluorescent light fixtures containing four 15-W fluorescent bulbs (UVP Inc., Upland, CA, USA) emitting UVB and UVA light from 280 to 400 nm (for emission spectrum see Figure 2 in section 1.2.). The utilized borosilicate glass Kimble KG-33, as well as the PFA foil used for the lining, are fully transparent in this wavelength range. The light bulbs are in close proximity (<2 cm) to the reactor tube to maximize the light influx. As the light bulbs create a substantial amount of heat, the cooling system of the reactor has been adapted to provide more cooling power. Several feet of copper tubing form two cooling loops inside the top-mounted UV lamps, removing the additional heat.

The entire cooling system was designed to provide two settings with different cooling intensity.

The first, lower setting, keeps the temperature constant, when the UV lamps are off. The second setting decreases the bath temperature of chiller #2 and increases the flow rate of the cooling air, as well as the flow rate of the coolant through the copper tubing to offset the heat generation, when the UV lamps are turned on. For photooxidation experiments, the minimum temperature achievable is -5°C due to the heat input from the lamps.

Two types of photooxidation experiments were conducted. For the qualitative assessment of the PAH photooxidation products, a saturated or half-saturated aqueous solution of the PAH was loaded onto the glass slide and subjected for a defined period of time to UVB light without any gas flow. For the HPLC analysis of the PAH-photooxidation products, the sample was obtained by opening the reactor and collecting the liquid on top of the glass slide.

68 For photooxidation experiments of 9H-fluorene, which incorporated a kinetic assessment of the reaction, the reactor was loaded with pure water and saturated with 9H-fluorene by the gas stream for a minimum of five hours. The photolysis was started by turning on the lamps inside the enclosure and switching the cooling system of the enclosure to the higher cooling capacity.

After a defined amount of time (0-180 min), the experiment was stopped, and the liquid phase sampled, by opening the reactor and washing the content of the glass slide with acetonitrile into a

10-mL volumetric flask. During the photooxidation experiment, the gas phase was also sampled by absorbing the gas stream into 30 mL of ice-cooled acetonitrile.

3.2. Bulk-Phase Reactor

The thin-film reactor is fairly complex and requires a significant amount of time to prepare for a new compound. To allow for a quick qualitative analysis of PAH photooxidation mixtures, a second reactor was constructed that allows the photolysis of PAH in the bulk phase.

Figure 31 shows a cross-sectional view of the reactor. The reactor consists of a light fixture identical to the ones used in the thin-film reactor. The light fixture is placed inside an open aluminum box, facing upward. Similar to the thin-film reactor, the light fixture is cooled through copper tubing through which a coolant coming from a chiller (Haake D1, Thermo Haake,

Karlsruhe, Germany) circulates. The cooling is necessary to avoid a temperature increase due to the heat generation of the UVB lamps. Dry, pressurized air is supplied to the box to avoid condensation of humidity inside the box or on the electronics.

The aqueous solution of the PAH is kept in 40-mL UVB-transparent borosilicate vials

(Qorpak, Bridgeville, PA, USA), which are held in place by a retaining bracket at the top of the box. To ensure homogeneity during the reaction, the solution is stirred with a micro-stir bar and a

69 magnetic stirrer (VWR submersible stirrer, VWR International, Radnor, PA, USA), which is mounted to the side of the box.

A typical experiment consists of loading the reactor with a vial, filled with an aqueous solution of the PAH in question and a magnetic stir bar. To initiate the photoreaction, the UV lamps and the cooling system are turned on. After a certain period of time the vial is removed from the reactor and the solution analyzed via HPLC.

Clear 40 mL Vial Retaining bracket Magnetic stir bar

Magnetic stirrer

2 UVB Lamps (ea 15 W) with external cooling

Aluminum housing

Figure 31. Cross-sectional view of the bulk-phase photoreactor

3.3. Sample Analysis

Three types of samples originate from the bulk-phase reactor and the thin-film reactor, corresponding to the different types of experiments conducted:

Qualitative experiments from the bulk phase reactor or the thin-film reactor (without gas flow) yield aqueous samples, which were analyzed via HPLC-UV/DAD-APPI/MS with and without the online concentration system.

The quantitative analysis of the solutions obtained from the 9H-fluorene photooxidation for the kinetic assessment is conducted via HPLC-UV/DAD-FLD with the online concentration

70 system. The obtained liquid- and gas-phase samples consisted predominantly of acetonitrile

(>50%), which prevents the direct application of the online concentration method, as the solid- phase extraction column cannot retain all compounds in mixtures with more than 10% acetonitrile. To make use of the online concentration system, these solutions were diluted by a factor of 10 with LC-MS grade water, so that the acetonitrile constitutes no more than 10% of the solution.

71 4. PHOTOOXIDATION PRODUCTS OF POLYCYCLIC AROMATIC HYDROCARBONS

In this chapter, the photooxidation products of several of the volatile PAH were analyzed with the methodology developed in Chapter 2. The goal was to provide a qualitative assessment of the photooxidation of volatile PAH and to identify known as well as unknown oxidation products of photodegradation of volatile PAH in the presence of water and air. The tested PAH included naphthalene, phenanthrene, pyrene, acenaphthene and 9H-fluorene.

4.1. Naphthalene

Two naphthalene photooxidation samples obtained from the thin-film reactor71 were analyzed. The samples were generated by saturating a film of 22 µm (thin-film) and 450 µm thickness (thick-film) with naphthalene and irradiating the saturated films for 16 and 17 hours respectively with UVB light, while keeping the thin-film reactor saturated with naphthalene.

Naphthalene is the most volatile of all PAH and also exhibits the highest water solubility of all tested PAH. As such the analysis of naphthalene is rather straight forward, not requiring any concentration step. An injection volume of 100 µL was sufficient for qualitative analysis.

Both samples were analyzed with the HPLC method listed as method C in Appendix A.

Figure 32 shows the chromatogram obtained from the HPLC-UV/DAD analysis of the thick-film

(450 µm) sample, and Figure 33 shows the chromatogram obtained from the HPLC-UV/DAD analysis of the thin-film (22 µm) experiment. The HPLC traces of the two experiments are nearly identical. In both samples, several of the naphthalene degradation products described in the

72 literature39,71 could be identified by matching their UV spectra (UV spectral matches see

Appendix D) and retention times with those of reference standards.

250

200

150

100

50

Average UV Absorbance 252-256 nm [mAU] 252-256 UV Absorbance Average 0

0 10 20 30 40 50 60

HPLC Retention Time [min] Figure 32. HPLC-chromatogram of naphthalene photooxidation mixture from the UVB photooxidation of naphthalene adsorbed and dissolved in 450-µm thick-water film

The identified photooxidation products are phthalide (3H-isobenzofuran-1-one), coumarin (2H-chromen-2-one), 1-indanone, 1,4-naphthoquinone and naphthalene-1-ol. As both samples show the same product distribution, it is evident that the chemical degradation pathways do not depend on the film thickness. All of the observed products are oxidized degradation products. Products larger than naphthalene, which could have formed from condensation, were not observed. Most of the products found can be traced back to the reaction of naphthalene with molecular oxygen and singlet oxygen as major oxidant.71

73 350

300

250

200

150

100

50

Average UV Absorbance 252-256 nm [mAU] 252-256 UV Absorbance Average 0

0 10 20 30 40 50 60

HPLC Retention Time [min] Figure 33. HPLC-chromatogram of naphthalene photooxidation mixture from the UVB photooxidation of naphthalene adsorbed and dissolved in 22-µm thick water film

4.2. Phenanthrene

Phenanthrene, with its low water solubility, requires the use of the online concentration system for the analysis of its oxidation products. Figure 34 shows the HPLC chromatogram of a phenanthrene photooxidation sample, which was obtained by injecting 2.5 mL with the online concentration system (HPLC-method C in Appendix A). The sample was obtained by UVB irradiation of a phenanthrene-saturated, 515-µm thick-water film for 12 hours in the thin-film reactor.73 9,10-Phenanthrenequinone, 6H-dibenzo[b,d]pyrane-6-one, 9H-fluorene-9-one and 9H- fluorene could be identified by their UV spectra (UV spectral matches see Appendix D) and by

74 the presence of the correct ions in their mass spectra. The oxidation products found match those described in the literature.66,145

3000

2500

2000

1500

1000

500 Unknown A

Average UV Absorbance 252-256 nm [mAU] 252-256 UV Absorbance Average 0 0 10 20 30 40 50 60 HPLC Retention Time [min]

Figure 34. HPLC chromatogram from a photooxidation mixture of phenanthrene obtained from UVB irradiation of a phenanthrene-saturated 515-µm thick-water film

9H-fluorene has been described as a phenanthrene-degradation product in the literature;66,145-146 however, its formation is somewhat surprising, as it seems unlikely that in an oxidative environment a fully reduced PAH would form. A more likely source of the 9H- fluorene (than the photochemical reaction) is its presence as an impurity in the phenanthrene utilized. For comparison, a phenanthrene photodegradation sample obtained from the photolysis of highly purified, sublimed phenanthrene (>99.5%, Sigma Aldrich, Milwaukee, WI, USA) in the bulk reactor was analyzed with the same method. Figure 35 shows the HPLC chromatogram

75 of this reaction mixture. The absence of 9H-fluorene and the significantly lower prevalence

(1/200 compared to the reaction products of the less pure phenanthrene) of 9H-fluorene-9-one

(the main oxidation product of 9H-fluorene) is striking. The 9H-fluorene-9-one found among the reaction products of the pure (99.5%) phenanthrene amounts to less than 0.29% of the initial phenanthrene concentration. Therefore it is clear that the previously found 9H-fluorene as well as the majority of the 9H-fluorene-9-one originate from the impurity of the utilized phenanthrene.

3000

2500

2000

1500

1000

500 Unknown A

Average UV Absorbance 252-256 nm [mAU] 252-256 UV Absorbance Average 0 0 10 20 30 40 50 60 HPLC Retention Time [min]

Figure 35. HPLC chromatogram of a photooxidation mixture of very pure phenanthrene (>99.5%) obtained from 12 hours UVB irradiation of a saturated phenanthrene solution in the bulk-phase reactor.

Even though the phenanthrene utilized for the photooxidation sample in Figure 34 was of high purity (>96%), it is essential to recognize that with a delivery method based on vapor pressure, a discrimination of the less volatile phenanthrene takes place, enriching the gas stream

76 with 9H-fluorene. All the following results in this section were therefore based on highly purified PAH, with a maximum of 1 ppm observable impurities. The purification procedures for each compound are described in Appendix C. The oxidation of the fluorene- and anthracene-free phenanthrene also did not yield any 9H-fluorene, but it did produce 9H-fluorene-9-one as photooxidation product.

6H-dibenzo[b,d]pyran-6-one, 9,10-phenanthrenequinone and 9H-fluorene-9-one were found in both chromatograms and can be therefore attributed to the photooxidation of phenanthrene. In both samples, an unknown, marked as ―Unknown A‖ in Figure 34 and Figure

35, was found. Figure 36 shows the UV and the APPI mass spectra of unknown A. The UV spectrum in Figure 36a is rather unspectacular, and does not provide any significant clues for the identification of unknown A, except that it is most likely an oxygenated compound.

100 a) Unknown A b) 223 225 80 60 40

Absorbance 20 197 195 0 200 300 400 500 600 [%] Ion Abundance Relative 180 200 220 240 260 m/z m/z

Figure 36. a) UV spectrum of unknown A and b) mass spectrum of unknown A

The dopant-assisted APPI-mass spectrum in Figure 36b yields significantly more information: Two positive ions at m/z 225 and m/z 197, as well as negative ions at m/z 223 and at m/z 195 can be identified. The molecular weight of 224 Da can be derived from the largest ions, as only even numbers for the molecular weight are possible, due to the elemental limitation to only carbon, hydrogen and oxygen. By utilizing the NIST isotope calculator for a molecular

77 mass of 224 Da with carbon, hydrogen and oxygen as constituting elements, a total of 19 permutations of elemental compositions can be found. Elemental compositions with more than

14 carbons were excluded, as it is unlikely for the utilized phenanthrene to form new carbon- carbon bonds during photooxidation. Elemental compositions with less than 13 carbons were also excluded, as most of these permutations would have required the breaking of more than one aromatic ring due to their high hydrogen content (to sustain the molecular mass despite the ―lost‖ carbon). The most likely compositions were C13H20O3, C14H8O3 and C14H24O2. The first formula,

C13H20O3, contains too much hydrogen to originate from a three-ring PAH, which is mirrored in its rings-plus-double-bonds value (RDB) of four. A RDB of four would have allowed only for one aromatic ring in the compound, which is unlikely to occur for a photooxidation product of phenanthrene, a three-ring compound with a RDB of 10. The other formula, C14H24O2 can also be eliminated because of its low RDB number of three and its high hydrogen content. C14H8O3

(RDB 11) is a clear choice, as the parent-PAH phenanthrene (C14H10 - RDB 10) is comparable in carbon and hydrogen content and also the RDB number.

The three oxygen atoms found in the elemental composition of unknown A, could allow for up to three different functional groups in the molecule. The utilization of the dopant-assisted

APPI-MS rules developed in section 2.6., seems therefore to be ideal to shed more light on the identity of unknown A. The observed ionization pattern in Figure 36b is quite interesting, as it does not fit directly to any of the tested compound classes. The negative [M-CO-H] ion at m/z 195 has not been observed before and cannot be assigned to any previously tested compound class. It is either a special case for this molecule, hinting at a carbonyl functionality, or a coeluting impurity. The negative [M-H] ion at m/z 223 indicates either a phenolic alcohol or a carboxylic acid. As the carboxylic acid would show an intense negative [M-COOH] ion at m/z

78 179, the presence of an acid can be ruled out. Therefore it seems that the unknown A contains at least one phenolic hydroxyl group. It is also clear that there is only one hydroxyl group in unknown A, as multiple hydroxyl functionalities on a molecule would result in the occurrence of additional [M-(x H)] ions.

Two more oxygen atoms are therefore available for different functional groups. The positive [M+H] ion at m/z 225 and the minor [M-CO+H] fragment at m/z 197 indicate the presence of either an ester or an aromatic ketone. With the previously found hydroxyl group as

RDB-neutral oxidation, a quinoid structure is more likely than an ester or simple ketone functionality, as only a quinone could yield an increase in the RDB number. Therefore, a hydroxyl-substituted quinone would yield the correct elemental composition and also provide the correct functionalities to fulfill the positive and negative mass spectrum of the unknown A.

There are a total of fifteen phenanthrenequinones possible, with only a few synthesized,147 of which 9,10-phenanthrenequinone is the most stable compound. As 9,10- phenanthrenequinone was also found as oxidation product of phenanthrene, it is conceivable that the unknown A is a derivative of 9,10-phenanthrenequinone. With the hydroxyl-substitution at the 1-, 2-, 3-, or 4-position, a total of four hydroxy-9,10-phenanthrenequinones isomers are possible. These compounds are commercially not available, and only for two, 2- and 3- hydroxyphenanthrenequinone, a complete, multi-step synthetic procedure is described in the literature.148-149

To obtain pure reference compounds for comparison, both 2- and 3-hydroxy-9,10- phenanthrenequinone were synthesized and purified in a 5-step synthesis described in

Appendix E. Both compounds were subjected to the same HPLC analysis as the phenanthrene- photooxidation samples. 3-hydroxy-9,10-phenanthrenequinone exhibited a mass spectrum

79 identical to that of unknown A; however, its UV spectrum is significantly different from the unknown’s UV spectrum. Figure 37 shows the mass spectrum and the UV spectral comparison of 2-hydroxy-9,10-phenanthrenequinone and unknown A. The comparison of Figure 37a with

Figure 36b shows that 2-hydroxy-9,10-phenanthrenquinone also exhibits the same mass spectral pattern as unknown A, except for the missing negative ion at m/z 195 (most likely a coeluting impurity). Additionally, the UV spectrum of the synthesized compound matches the UV spectrum of unknown A, and the HPLC-retention time of the synthesized compound is identical with the one of unknown A. Therefore, the identity of unknown A is unequivocally established as 2-hydroxy-9,10-phenanthrenequinone.

100 a) 225 [M+H]+ b) 80

60 Synthesized 2-hydroxy-9,10-phenanthrene 40 223 [M-H]- Absorbance Unknown A 20

Relative Ion Abundance 197 [M+H-CO]+

0 160 180 200 220 240 200 300 400 500 600 m/z m/z

Figure 37. a) APPI-MS mass spectrum of synthesized 2-hydroxy-9,10-phenanthrenequinone and b) UV spectra of unknown A and synthesized 2-hydroxy-9,10-phenanthrenequinone

Hydroxy-9,10-phenanthrenequinones have been suspected to form from oxidation of

9,10-phenanthrenequinone with hydroxyl radicals in the gas phase.150 The formation of 2- hydroxy-9,10-phenanthrenequinone in this work proceeds most likely from 9,10- phenanthrenequinone, as 2-hydroxyphenanthrene, found during hydroxyl radical oxidation of phenanthrene,150 was not present among the oxidation products. Therefore the introduction of a

80 hydroxyl group into 9,10-phenanthrenequinone via oxidation by molecular oxygen is the more probable reaction mechanism.

4.3. Pyrene

Pyrene photooxidation has been investigated,42 and 1,6-pyrenequinone, 1,8- pyrenequinone and 1-hydroxypyrene were identified as the main degradation products. The main reaction path is thought to proceed via oxidation with molecular oxygen42 to 1-hydroxypyrene, which reacts further to the two pyrenequinones. The analysis of pyrene, the largest and least water-soluble volatile PAH, and its photooxidation products requires the use of the online concentration system. Figure 38 shows the HPLC chromatogram (obtained with method C in

Appendix A) of a half-saturated pyrene solution, which has been subjected to 12 hours UVB irradiation in the bulk-phase reactor. The chromatogram was obtained by using the online concentration system with a 5-mL injection. All of the oxidation products reported in literature were found in this sample by comparing their UV spectra with those of reference substances or those available in the literature. Apart from these known photooxidation products, an additional compound was found: 4,5-pyrenequinone was identified by matching its UV spectrum with that of a commercially-available reference standard, as can be seen in Figure 39.

The mechanism of the photochemical degradation of pyrene is suspected to proceed by reaction with molecular oxygen.42 Hydroxyl radicals in the tested system seem to play in the tested system a minor role, as the expected product of the hydroxyl radical reaction,151 1-hydroxy- pyrene, is only found in minor quantities and other hydroxy-pyrenes are also missing. Sigman et al.42 described the mechanism of pyrene photooxidation in water with UVB light as an initial oxidation step of pyrene with molecular oxygen to 1-hydroxypyrene, with subsequent

81

1200

1000

800

600

400

200

Average UV Absorbance 252-256 nm [mAU] 252-256 UV Absorbance Average 0

0 10 20 30 40 50 60 HPLC Retention Time [min]

Figure 38. HPLC chromatogram of the photooxidation mixture of a half-saturated pyrene mixture after 12 hours UVB irradiation.

formation of 1,6- and 1,8-pyrenequinone. Although this is most likely the major reaction pathway, it does not account for the formation of 4,5-pyrenequinone, which is the typical reaction product of the reaction of singlet oxygen at the 4,5-position.152 Sigman et al. indicated that reaction with singlet oxygen was not observed in their system, but it is more likely, that 4,5- pyrenequinone was not observed due to incomplete chromatographic separation, as this work and the work of Sigman et al. used similar conditions (UVB light, water bulk solution) and the 4,5- pyrenequinone coelutes with 1,8-pyrenequinone.

82 4,5-Pyrenequinone

Pyrene photooxidation

Absorbance product

200 300 400 500 600

m/z

Figure 39. UV spectra of 4,5-pyrenequinone and pyrene photooxidation product identified as 4,5-pyrenequinone

4.4. Acenaphthene

The photodegradation of acenaphthene in pure water has not been described in the literature before; literature-available work focuses on the kinetic degradation of acenaphthene,46,55,153 especially with advanced oxidation techniques such as ozonolysis153 or oxidation with hydrogen peroxide.55 The only acenaphthene-photolysis study detailing reaction products was conducted by Reyes at al.154 In their study they described the photolysis of acenaphthene adsorbed on silica, which yielded 1-acenaphthenone, 1-acenaphthenol and 1,8- naphthalenedicarboxaldehyde as products.

83 Figure 40 shows the HPLC chromatogram (HPLC method D in Appendix A) from a 500-

µL injection onto the online concentration system of a half-saturated, aqueous acenaphthene solution, which has been subjected to UVB irradiation in the bulk reactor for 150 minutes. Two products, which could be identified by their UV spectra (spectral matches see Appendix D), are

1-acenaphthenone and 1-acenapthenol. As can be seen in Figure 40, there is quite an abundance of compounds (peaks), indicating a fairly complex chemistry.

140

120

100

80

60

40

20

0

Average UV Absorbance 252-256 nm [mAU] 252-256 UV Absorbance Average

0 10 20 30 40

HPLC Retention Time [min]

Figure 40. HPLC chromatogram of the photooxidation mixture of an aqueous, half-saturated acenaphthene solution after 12 hours UVB irradiation, 500-µL injection volume

The photodegradation of acenaphthene is thought to be initiated by molecular oxygen attack at the saturated cyclopenta-ring, allowing the ring to open.154 Once this ring is activated, the further degradation is centered on the transformation at the saturated ring, without affecting

84 the aromatic rings. The biogenic degradation of acenaphthene follows a similar pathway of degradation.155 The degradation with oxygen is substantially different from the oxidation via ozone, which allows for the breakage of the aromatic body, forming substituted indanones.153,156

4.5. 9H-Fluorene

9H-Fluorene photooxidation in pure water has been discussed in the literature, and 9H- fluorene-9-one, 9H-fluorene-9-ol and 9H-fluorene-2-ol have been reported as photooxidation products.43,45,49,63,157 Figure 41 shows the HPLC chromatogram (5-mL injection volume, HPLC method D in Appendix A) of an aqueous, half-saturated 9H-fluorene solution subjected to UVB irradiation in the thin-film reactor for 180 min at 20°C and at a film thickness of 369 µm. The major product of 9H-fluorene photooxidation is clearly 9H-fluorene-9-one, which was identified by its UV and mass spectra. 9H-fluorene-9-ol was also identified as an oxidation product, whereas the previously reported 9H-fluorene-2-ol63 could not be found in the reaction mixture.

A so-far unknown product, unknown B, elutes before 9H-fluorene-9-ol in Figure 41.

Figure 42 shows the UV spectrum of unknown B, which exhibits the typical feature-less UV spectrum of an oxygenated compound. Attempts to match the UV spectrum to those of available standards, such as 1-hydroxy-9H-fluorene-9-one or 1-hydroxy-9H-fluorene-9-one (suspected candidates because of the previous identification of 2-hydroxy-9,10-phenanthrenequinone) or to those available in the literature for 9H-fluorene-1-ol and 9H-fluorene-2-ol failed, and indicated the novelty of this product.

Figure 43 shows the positive and the negative mass spectrum of the unknown B. The positive mass spectrum consists of the major positive ion at m/z 181 and two more positive ions

85 at m/z 198 and 197. The negative mass spectrum shows two ions at m/z 197 and 169. The mass spectrum is peculiar, as it does not fit easily within the rules established in the previous chapter.

3000

2500

2000

1500

1000

500

Unknown B

Average UV Absorbance 252-256 nm [mAU] 252-256 UV Absorbance Average 0

0 10 20 30 40 50 60 HPLC Retention Time [min]

Figure 41. HPLC chromatogram from a photooxidation mixture of 9H-fluorene obtained from UVB irradiation of an aqueous, half-saturated 9H-fluorene 369-µm thick water film

The molecular weight can be determined to be 198 Da, based on the negative [M-H] ion at m/z 197 and the positive [M] ion at m/z 198. The most likely molecular formula was derived with the NIST isotope calculator as C13H10O2 with a RDB number of nine.

The positive ion mass spectrum in Figure 43a shows [M] and [M-H] ions at m/z 198 and

197, as well as a [M-OH] fragment at m/z 181, indicating a non-phenolic alcohol, similar to 9H- fluorene-9-ol. In contradiction to this preliminary conclusion is the existence of a negative ion

86 spectrum, as an aliphatic alcohol is not expected to exhibit one. The negative [M-H] ion at m/z

197 is normally indicative of a phenolic hydroxyl-group, and a dihydroxyl-substituted fluorene would fit the molecular formula of C13H10O2. The key ion excluding such a dihydroxy- compound, is the negative [M-CO-H] ion at m/z 169, which could not be observed by a hydroxyl-substituted 9H-fluorene-9-ol. Furthermore, any dihydroxyfluorene would be more polar than 9H-fluorene-9-ol, and therefore would elute much earlier than unknown B in Figure

41. Therefore unknown B is not a dihydroxyfluorene, but rather constitutes a yet undescribed compound class.

Unknown B

Absorbance

200 300 400 500 600

m/z

Figure 42. UV absorbance spectrum of unknown B

87 100 10 a) 181 b) 197 8 80 169 6 60 4 40 2 20 197 198 0 0 Relative Ion Abundance [%] Ion Abundance Relative 140 160 180 200 220 [%] Ion Abundance Relative 140 160 180 200 220 m/z m/z

Figure 43. a) positive and b) negative APPI-MS mass spectrum of unknown B

The compound exhibits a negative [M-CO-H] ion at m/z 169, indicative of a CO- abstraction. A CO-abstraction in the negative ion spectrum has not been observed before for any of the tested compound classes, even those containing a carbonyl functionality. Nonetheless, the negative [M-CO-H] ion is still pointing toward a carbonyl functionality, as only ketones and aldehydes have shown the capability of CO-abstraction (though in the positive mass spectrum).

The calculated hydrogen deficiency, based on the most likely elemental composition

C13H10O2, requires nine rings or double-bonds. As the carbonyl functionality accounts for one double bond, eight additional rings or double bonds have to be encountered. As only a few oxidation products are observed, it is reasonable to assume that the aromatic bonds of 9H- fluorene have not been broken, but that instead the carbon-carbon bond connecting the two phenyl rings at the 9-position or at the 4a/b-position of 9H-fluorene have been severed. With the identified building blocks of one hydroxyl group, one carbonyl group and two phenyl rings, only a few molecular arrangements will yield a molecular weight of 198 Da.

Figure 44 shows some of the most likely candidates. They all posses a molecular weight of 198 Da and consist of one hydroxyl group, one carbonyl group and two phenyl rings. The first

88 compound, in Figure 44 2-phenylbenzoic acid is a carboxylic acid and can be ruled out as a candidate, as the typical [M-45] ion for carboxylic acids was not found in the negative mass spectrum of unknown B. The second compound, 2-hydroxybenzophenone, is also an unlikely candidate, as it represents a phenol and a ketone (similar to the 2-hydroxy-9,10- phenanthrenequinone identified in section 3.2) for which the observed CO abstraction in the negative mass spectrum and the positive [M-1] ion would not be expected.

Figure 44. Potential candidates for unknown B, containing two phenyl rings, one carbonyl and one hydroxyl group, with elemental composition C13H10O2.

The third compound, 2'-hydroxy-1,1'-biphenyl-2-carboxaldehyde, and the fourth compound, 6H-dibenzo[b,d]pyran-6-ol, seem to be more suitable candidates. The aldehyde is expected to yield a positive [M-17] ion and would possess a CO group, which could be abstracted to yield the negative [M-CO-H] ion. Its phenolic hydroxyl group would also explain the negative [M-1] ion in unknown B’s negative mass spectrum.

The fourth compound in Figure 44, 6H-dibenzo[b,d]pyran-6-ol, has an aliphatic alcohol group, which would yield the correct positive mass spectrum with [M], [M-H] and [M-OH] ions.

This seemingly complementary behavior of 6H-dibenzo[b,d]pyran-6-ol and 2'-hydroxy-1,1'- biphenyl-2-carboxaldehyde is not a coincidence. Both compounds can transform into each other via tautomeric rearrangement. 6H-dibenzo[b,d]pyran-6-ol is the internally formed semi-acetal of the 2'-hydroxy-1,1'-biphenyl-2-carboxaldehyde.158 Due to this tautomerism, the semi-acetal’s

89 behavior during electron-impact-ionization mass spectrometry has been subject to detailed studies.158-159 Figure 44 shows the ionization mechanism, as it was suggested by Sollazzo et al.,159 for electron-impact ionization.

Figure 45. Ionization mechanism of 6H-dibenzo[b,d]pyran-6-ol during electron-impact ionization

Several of the positive ions (m/z 198, 197, 181) portrayed in Figure 45 can be directly related to the positive APPI mass spectral signal in Figure 43. Based on the electron-impact ionization mechanism, Figure 46 depicts the most likely ionization mechanism of 6H- dibenzo[b,d]pyran-6-ol for dopant-assisted APPI-MS.

6H-Dibenzo[b,d]pyran-6-ol fulfills therefore all requirements derived from unknown B:

6H-Dibenzo[b,d]pyran-6-ol possesses the required elemental composition of C13H10O2, and it provides a conclusive mechanism for the observed ions. It seems that with all this abundance of

90 mass spectral evidence, a synthesis of 6H-dibenzo[b,d]pyran-6-ol would not be necessary.

However, as the APPI-MS is a photochemical reaction chamber, in which a tautomeric rearrangement from the free aldehyde to the semi-acetal could occur, it is not clear, from the mass spectrum alone, if either the aldehyde or the semi-acetal was observed. Therefore the 6H- dibenzo[b,d]pyran-6-ol was synthesized as a pure compound (details see Appendix E) and analyzed with HPLC-UV/DAD-APPI/MS for comparison with the fluorene oxidation sample.

Figure 46. Suggested ionization mechanism for 6H-dibenzo[b,d]pyran-6-ol in dopant-assisted APPI-MS. Negative ions in blue; positive ions in red

91 Figure 47 shows the APPI mass spectrum of the synthesized 6H-dibenzo[b,d]pyran-6-ol and the UV spectral match of the synthesized 6H-dibenzo[b,d]pyran-6-ol with the spectrum of unknown B. In addition to the identical MS spectrum and the excellent UV spectral match, the observed retention time of the synthesized 6H-dibenzo[b,d]pyran-6-ol was identical to that of unknown B, unequivocally confirming the identity of unknown B as 6H-dibenzo[b,d]pyran-6-ol.

100 a) b) 181 [M-17]+ 80

60 Synthesized 40 6H-dibenzo[b,d]pyran-6-ol

Absorbance + Unknown B 20 198 [M] Relative Ion Abundance 169 [M-29]- 197 [M-H]- 0 160 180 200 220 240 200 300 400 500 600 m/z m/z

Figure 47. a) Mass spectrum of synthesized 6H-dibenzo[b,d]pyran-6-ol and b) UV spectra of unknown B and synthesized 6H-dibenzo[b,d]pyran-6-ol

Although 6H-dibenzo[b,d]pyran-6-ol is not the dominant oxidation product, it is significant, as contrary to the photostable 9H-fluorene-9-one, it can react in alkaline medium to the aldehyde, which can then oxidize further. Additionally, 6H-dibenzo[b,d]pyran-6-ol is used as structural backbone for active pharmacological ingredients,160-161 which would insinuate potentially significant health effects of this 9H-fluorene-photooxidation product. 6H- dibenzo[b,d]pyran-6-ol is also known to dimerize easily to 6,6’-oxybis-6H-dibenzo[b,d]pyran upon heating;162-163 it might therefore have the ability to lead to larger molecules, when 9H- fluorene-containing fog droplets dry up in the sun.

92 4.6. Summary

In this chapter, naphthalene, pyrene, phenanthrene, acenaphthene, and 9H-fluorene were subjected to UVB light in pure water in the presence of air, and their reaction mixtures were analyzed with the analytical techniques introduced in Chapter 2. Based on these results several conclusions can be drawn.

The purity of the utilized PAH for photooxidation studies is of significant importance, especially, when the PAH is delivered by techniques that allow for partitioning. Impure PAH can result in misidentification of products and the definition of false reaction pathways.

Sufficient chromatographic separation is also a key parameter for identifications, demonstrated in the case of 4,5-pyrenequinone. 4,5-Pyrenequinone has been overlooked as a reaction product in a previous study42 of pyrene photooxidation, leading the authors to the omission of the reaction with singlet oxygen as pathway for pyrene photodegradation.

The rules for the interpretation of dopant-assisted APPI mass spectra, developed in Chapter

2, have been successfully tested on complex unknown compounds such as 2-hydroxy-9,10- phenanthrenequinone and 6H-dibenzo[b,d]pyran-6-ol. The developed method allows for a targeted synthesis and/or acquisition of reference substances. In case of the successful identification of 6H-dibenzo[b,d]pyran-6-ol, it prevented the tedious synthesis and testing of all twenty dihydroxy-9H-fluorene isomers and led quickly to the correct compound. The utilization of the dopant-assisted APPI mass spectrometry has therefore enabled the successful identification of 2-hydroxy-9,10-phenanthrenequinone and 6H-dibenzo[b,d]pyran-6-ol for the first time as photooxidation products of phenanthrene and 9H-fluorene, respectively.

Although the tested PAH are quite different in their chemical structures, they exhibit a common oxidation mechanism. By reviewing all the observed UVB photooxidation products of

93 the different volatile PAH, it can be deduced that the dominant degradation pathways of the PAH are the reaction with molecular oxygen and singlet oxygen. Hydroxyl-radical formation and its chemical attack are less likely to occur. If the reaction with hydroxyl radicals would have been significant, in the case of phenanthrene, all of the hydroxyphenanthrenes would have been observed.150

Compounds such as 6H-dibenzo[b,d]pyran-6-one can be synthesized via Baeyer-Villiger oxidation,164 which utilizes peroxy acids as reagents. Similar peroxy compounds can be formed through oxygen addition to the PAH47 and might be the ―reagent‖ for the oxidation in the thin film.

The degradation through singlet oxygen is confirmed by the presence of 4,5-pyrenequinone and 1,4-naphthoquinone, which are both typical products of the reaction of pyrene and naphthalene, respectively, with singlet oxygen.

The addition of other atmospheric species will change the degradation chemistry of PAH dramatically. In the case of acenaphthene, the addition of ozone would allow the opening of the aromatic ring to indanones, and in the case of phenanthrene the addition of hydroxyl-radicals would yield all five hydroxyphenanthrenes, and therefore most likely all four hydroxy-9,10- phenanthrenequinones.

94 5. PHOTOOXIDATION OF 9H-FLUORENE IN THIN WATER FILMS

Air-water interfaces are abundant in the environment, and they provide a unique venue for the photochemical transformation of PAH, due to the ability of PAH to adsorb to the interface.70-73,83-

84,92,165

5.1. Uptake of 9H-Fluorene onto Water Films

The air-water interface is a very thin layer, only a few Ångstrom thick,166 in which the density gradually changes from bulk water to air. PAH find in this layer a free energy minimum, allowing them to accumulate at the interface.81

This interfacial accumulation or adsorption can be thermodynamically described by introducing the interface as a separate phase.167 Figure 48 shows the three phases, bulk water, bulk air and the air-water interface, and the thermodynamic relations linking their concentrations via their partitioning coefficients. The partitioning between the liquid (aqueous) phase and the gas phase is described by Henry’s law, where a compound in the gas phase is in equilibrium with its liquid solution. A similar relation describes the equilibrium between the air-water interface with the bulk-liquid phase. The partitioning coefficient between the interface and the bulk-liquid phase or the air-bulk phase is described by the ratio of the surface concentration to the bulk concentrations. Although the interface, the steep gradual change from the liquid to the gas phase, occupies actual volume, it is usually approximated as an area, with a surface concentration ΓS instead of a volumetric concentration. As the interface (10-10 m range) is many orders of

95 magnitude thinner than any liquid film (10-6 m range), this approximation is more than sufficient for the current work, which deals with thin films ranging from 22 to 515 µm.

Air

cA

Henry’s Law Γ c K  S K  W IA WA cA cA

ΓS K IW  cW Water Interface

cW Γs

Figure 48. Thermodynamic equilibrium relationships between bulk air, bulk water and the air- water interface

The uptake of a compound by dissolution and adsorption into a thin-water film can be described with a simple mass balance over a prism-shaped thin film, with surface area A and film thickness δ. The total mass mT of the compound associated with the film is comprised of the mass dissolved in the bulk-water phase, mBW, and the mass adsorbed at the air-water interface, mIA.

mT  mBW  mIA Eqn. 1

The mass adsorbed at the air-water interface can be calculated by multiplying the surface concentration by the surface area (mIA = ΓS·A). By inserting the mass adsorbed at the interface

96 into Equation 1 and then dividing the entire equation by the total volume of the thin film

(V = A·δ), a relationship in the form of concentrations is obtained.

Γ cT  c0  S Eqn. 2 W W δ

T The total concentration cW in the thin film can be easily assessed by analyzing the concentration after dilution (increasing δ) e.g. with HPLC. The direct determination of the bulk-

0 water cw and the interfacial concentration (ΓS/δ) require in-situ measurements.

By utilizing the thermodynamic equilibrium relationships detailed in Figure 48, Equation

2 can be expressed in terms of the bulk-gas-phase concentration (ΓS = KIA·CA) and furthermore in terms of the bulk-water-phase concentration by applying (CA = CW/KWA. The resulting equations, Equation 3 to Equation 5, allow for some insight into the importance of interfacial adsorption for volatile compounds.

K cT  c  c  IA Eqn. 3 W W A δ

T cW K IA cW  cW   Eqn. 4 K WA δ

 K 1  T  IA  Eqn. 5 cW  cW 1    K WA δ 

When the film thickness δ is small or KIA is large, the mass accumulated at the interface dominates. When the film is thick, i.e. δ is large, the bulk water phase concentration is dominant, and the uptake can be described by Henry’s law. From Equation 4, another interesting observation regarding the type of compounds that will exhibit surface adsorption can be made:

For polar, volatile compounds with high solubility in water, i.e. large KWA, surface adsorption is of lesser importance than for non-polar compounds such as volatile PAH with small KWA and large cA.

97

5.2 Photooxidation of 9H-Fluorene in Thin-Water Films

Previous studies have shown that degradation of PAH such as naphthalene and phenanthrene is greatly enhanced at the air-water interface, allowing for up to ten-times faster degradation rates.72-73

The photooxidation of 9H-fluorene in thin-water films of 37-µm and 185-µm thickness was tested at 19.8°C in the thin-film reactor described in Chapter 3. The water films were saturated with 9H-fluorene for at least five hours prior to the start of the photooxidation. After the photoreaction was started, 9H-fluorene was continuously supplied during the reaction. After a defined period of time, the concentrations of the products formed in the liquid and gas phases were assessed by HPLC analysis (Method F in Appendix A).

Figure 49 shows the measured concentrations of 9H-fluorene, 9H-fluorene-9-one, 9H- fluorene-9-ol and 6H-dibenzo[b,d]pyran-6-ol versus the reaction time from photooxidation experiments of 9H-fluorene in a 185-µm thick water film. It is evident in Figure 49 that the 9H- fluorene concentration, depicted as open circles, changes only slightly from experiment to experiment (with time), and can be approximated as constant (indicated by the dashed line in

Figure 49). 9H-Fluorene-9-one, whose concentrations are depicted by the open squares in Figure

49, is the major photooxidation product and forms in concentrations comparable to 9H-fluorene

(scale at the left side). The yields of 9H-fluorene-9-ol (filled circles) and 6H-dibenzo[b,d]pyran-

6-ol (crosses in Figure 49) are considerably less (scale on the right side), identifying these species as minor products of the reaction.

98 5 2.0 185 µm 9H-Fluorene-9-ol 6H-Dibenzo[b,d]pyran-6-ol 4 9H-Fluorene-9-one 9H-Fluorene 1.5

mol/L] mol/L]

 3  1.0 2

0.5 Concentration [ Concentration 1 [ Concentration

0 0.0 0 25 50 75 100 125 150 175 200

Time [min]

Figure 49. Concentration of 9H-fluorene-9-ol, 6H-dibenzo[b,d]pyran-6-ol, 9H-fluorene-9-one and 9H-fluorene in the 185-µm film during photodegradation (lines fitted to Equation 11)

A similar set of photooxidation experiments was conducted at the same temperature of

19.8°C, but with a water film of only 37-µm thickness, increasing the surface-to-volume ratio by a factor five. Figure 50 shows the concentrations of the 9H-fluorene, 9H-fluorene-9-one, 9H- fluorene-9-ol and 6H-dibenzo[b,d]pyran-6-ol in the 37-µm film versus the reaction time. The

9H-fluorene concentration, depicted as the open circles in Figure 50, does not change significantly with time and can therefore be approximated as constant. As can be seen in Figure

50, 9H-fluorene-9-one (open squares) is the major product of 9H-fluorene oxidation, reaching

99 concentrations that surpass the concentration of the parent PAH. The other products, 9H- fluorene-9-ol (filled circles) and 6H-dibenzo[b,d]pyran-6-ol (crosses) also show enhanced yields relative to the experiments with the thicker water film. In order to quantify this enhancement, a suitable kinetic description has to be derived.

4 2.0 37 µm 9H-Fluorene-9-ol 6H-Dibenzo[b,d]pyran-6-ol 9H-Fluorene-9-one 3 9H-Fluorene 1.5

mol/L] mol/L]

 

2 1.0

1 0.5

Concentration [ Concentration [ Concentration

0 0.0 0 25 50 75 100 125 150 175 200

Time [min]

Figure 50. Concentration of 9H-fluorene-9-ol, 6H-dibenzo[b,d]pyran-6-ol, 9H-fluorene-9-one and 9H-fluorene in the 37-µm film during photodegradation (lines fitted to Equation 11)

The photochemical degradation of PAH is commonly assumed to follow first-order reaction kinetics.42,49,68 The integrated kinetic equation of a first order reaction for a batch reactor

(Equation 6) shows an exponential decay in the PAH concentration (cPAH) over time (t).

100 Consequentially, the formation of a photochemically stable OPAC, from

k1 PAH OPAC(stable), is described in Equation 7 by an exponential increase until a maximum is reached. The concentrations of non-stable photooxidation products, which either continue to react further or are withdrawn from the liquid phase by partitioning, exhibit an initial exponential increase with a later exponential decrease, depending on the rate constants of the reactions

PAH k'1  OPAC1 k'2 OPAC2. Equation 8 describes the concentration of the intermediate product, OPAC1.168

k1t cPAH (t)  cPAH (0)e Eqn. 6

k1t cOPAC(stable) (t)  cPAH(0)1 e  Eqn. 7

k'1 k'1 t k'2 t cOPAC1(t)  cPAH (0) e  e ; cOPAC1(0)  0 Eqn. 8 k'2 k'1

For the special case that the PAH concentration is kept constant through continuous supply of the PAH in the gas phase, these mathematical expressions simplify significantly. The formation of a stable product can then be described as a simple linear curve by Equation 9. The expression for a consecutive reaction or a reaction with loss via partitioning to the gas phase also simplifies greatly to Equation 10.

cOPAC,stable(t)  cPAH(0)k1  t Eqn. 9

k'1 k'2 t cOPAC1(t)  cPAH(0)1 e  Eqn. 10 k'2

The concentrations of all the products, 9H-fluorene-9-ol, 6H-dibenzo[b,d]pyran-6-ol and

9H-fluorene-9-one were fitted to Equation 10, to account for partitioning into the gas phase or further reaction. Gas-phase oxidation of 9H-fluorene was not observed without a water film

101 present. The kinetic evaluation of the data was based on the dimensionless concentrations, normalized by the individual 9H-fluorene concentration observed for each experiment.

Table 9 gives the calculated rate constants and the coefficient of determination R2 for the oxidation products formed in the 185-µm thick-film and for the 37-µm thin-film experiments.

The constant k1 denominates the growth of each species, whereas k2 is a composite of further degradation and the rate of mass transfer to the gas phase. The values in Table 9 demonstrate

Table 9. Kinetic constants for 9H-fluorene oxidation products formed in a 185-µm water film and a 37-µm water film at 19.8°C

185-µm Water Film 37-µm Water Film

2 2 Product k1 k2 R k1 k2 R [min-1] [min-1] [min-1] [min-1]

9H-Fluorene-9-one 5.801×10-3 8.839×10-3 0.9664 14.28×10-3 7.051×10-3 0.9852

9H-Fluorene-9-ol 6.184×10-4 4.316×10-3 0.9188 67.25×10-4 0.01619 0.9807

6H-Dibenzo[b,d]pyran-6-ol 5.290×10-4 1.774×10-3 0.9893 39.39×10-4 5.809×10-3 0.9723

clearly that the formation of 9H-fluorene oxidation products is greatly enhanced in thin-water films. The increase in the reaction rate constant k1 for 9H-fluorene-9-one, 148%, is substantial, whereas those for 9H-fluorene-9-ol and 6H-dibenzo[b,d]pyran-6-ol, 993% and 645%, respectively, are even higher. These substantial gains with the five-fold increase in the surface- to-volume ratio provide evidence that the interface becomes the dominant venue for reaction.

Although Table 9 shows that the k1-value for the formation of 9H-fluorene-9-one increases significantly, its k2-value stays roughly the same. As the k2-value describes further decomposition or transport out of the film, 9H-fluorene-9-one can be considered a stable photooxidation product that does not react further. In contrast, 9H-fluorene-9-ol’s k2-value rises

102 by 275%, and that of 6H-dibenzo[b,d]pyran-6-ol by 227%, indicating that both of these species are reactive intermediates.

5.3. Discussion

The photooxidation of 9H-fluorene in thin-water films in the presence of air yields three main products: 9H-fluorene-9-one, 9H-fluorene-9-ol and 6H-dibenzo[b,d]pyran-6-ol. When the surface-to-volume ratio of the thin film is increased, all products experience a significant increase in their formation rates, proving that the reaction predominantly occurs at the interface.

The mechanism of 9H-fluorene photodegradation has been explored169 only in parts and can be expanded to incorporate the current results. Figure 51 shows a schematic overview of the most likely reaction pathways. The oxidation of 9H-fluorene occurs exclusively at the 9-position with molecular oxygen forming 9H-hydroperoxyfluorene as reactive intermediate.169 This high selectivity of the reaction towards the 9-position of 9H-fluorene is due to the enhanced reactivity of the hydrogen atoms at this position, which is described by the known high acidity of 9H- fluorene.138 The peroxo compound formed can either rearrange to 6H-dibenzo[b,d]pyran-6-ol or split off water to yield 9H-fluorene-9-one. The latter reaction seems to be a dominant one, as 9H- fluorene-9-one is the main oxidation product of 9H-fluorene.

The formation of 9H-fluorene-9-ol, although seemingly an intermediate in the reaction to

9H-fluorene-9-one, is not that straight forward. The direct oxidation of 9H-fluorene to 9H- fluorene-9-ol would require the insertion of only one oxygen atom. This is an unlikely reaction step in this system (UVB, ambient) as single oxygen atoms could not form due to the substantially high energy required for the oxygen-oxygen bond scission of molecular oxygen

(119 kcal/mol),170 an amount of energy that is not available in this system. Oxygen atoms would

103 also exhibit significantly higher reactivity, resulting in the loss of the selectivity of the oxidation.

As in this work the only oxygenated compounds observed were those with the oxygen- containing substituent in the 9-position, oxidation of 9H-fluorene to 9H-fluorene-9-ol via oxygen atoms is very unlikely.

Results on the photolysis of other PAH, as well as the lack of hydroxy-substituted products (e.g. hydroxy fluorenones) from 9H-fluorene photooxidation indicate that the hydroxyl radical does not play a significant role in the photooxidation of 9H-fluorene when only water and oxygen are present. Therefore a radical chain reaction with hydroxyl radicals removing a hydrogen atom at the 9 position and subsequent chain reaction to 9H-fluorene-9-ol are also unlikely.

A potential explanation for the formation of 9H-fluorene-9-ol during the photooxidation of 9H-fluorene in the presence of water and air could be the oxidation of a 9H-fluorene molecule with 9H-hydroperoxyfluorene, yielding two 9H-fluorene-9-ol molecules.

Also non-trivial is the further oxidation of 9H-fluorene-9-ol to 9H-fluorene-9-one: direct oxidation of the alcohol with molecular oxygen would result in 9H-hydroperoxyfluorene-9-ol species. Further decomposition of the peroxo compound would require another molecule such as

9H-fluorene or another 9H-fluorene-9-ol to capture one of the oxygen atoms of the 9H- hydroperoxyfluorene-9-ol. There are two other pathways which should be considered: the oxidation of a 9H-fluorene-9-ol with 9H-hydroperoxyfluorene to one fluorenone and one fluoreneol molecule or the photo-initiated disproportioning of 9H-fluorene-9-ol to one 9H- fluorene-9-one and 9H-fluorene.

For the formation of 6H-dibenzo[b,d]pyran-6-ol there are two conceivable chemical pathways, either the photochemical rearrangement of the 9H-hydroperoxyfluorene or the

104 oxidation of 9H-fluorene-9-ol with 9H-hydroperoxyfluorene. Although the oxidation with peroxo compounds can lead to Baeyer-Villiger type oxidation,171 which inserts an oxygen next to a carbonyl functionality, it is unlikely that this type of oxidation occurs, as 6H- dibenzo[b,d]pyran-6-one, the product of Bayer-Villiger oxidation of 9H-fluorene-9-one, was not found. Therefore the most likely reaction for the formation of 6H-dibenzo[b,d]pyran-6-ol is the photochemical rearrangement of the peroxo compound.

These potential formation pathways discussed so far have to be considered to be only tentative reaction steps, as the data currently available do not allow for a complete deduction of the reaction mechanism. The reactions proposed above should serve therefore only as a potential starting basis for future work. The formation of 9H-hydroperoxyfluorene from 9H-fluorene is consistent with the observed products, and 9H-fluorene-9-one could be confirmed as a stable photooxidation product of 9H-fluorene photooxidation in the presence of water and air.

Figure 51. Possible reaction pathways for 9H-fluorene photooxidation

105 The photooxidation of 9H-fluorene leads to the generation of polar compounds, which due to their polarity are predominantly associated with the aqueous phase. Atmospheric photooxidation of 9H-fluorene at the interface of hydrometeors could therefore provide an atmospheric sink for 9H-fluorene, as the polar 9H-fluorene-9-one can accumulate in much higher concentrations than 9H-fluorene in the thin-water film. In the environment, 9H-fluorene in the gas phase would partition to a hydrometeor and react very quickly to a polar oxidation product at the interface. 9H-fluorene’s oxidation products would be taken up into the hydrometeor and therefore allow for the removal of 9H-fluorene from the atmospheric gas phase.

106 6. CONCLUSIONS

The purpose of this dissertation was the evaluation of the photooxidation of PAH with respect to their degradation products and the evaluation of the photodegradation of PAH at the air-water interface. The contributions to the scientific community presented in this thesis are summarized in three parts: advances made in the analytical methodology, the identification of newly identified UV oxidation products of PAH and the evaluation of the interfacial degradation of 9H-fluorene in thin-water films.

6.1. Advances in the Analytical Methodology of OPAC

The need for an advanced analytical technique for the analysis of photodegradation products of PAH prompted the development of several significant improvements of analytical hardware, in particular the online concentration system and the gas-phase dopant delivery system for dopant-assisted atmospheric APPI-MS.

The online concentration system described in Chapter 2 is a convenient and effective method to concentrate aqueous samples. It is superior to other extraction and concentration techniques in terms of recovery and speed, and it eliminates the use of toxic chemicals such as dichloromethane as extraction solvent. The demonstrated ability to concentrate samples by three orders of magnitude allows the analysis of challenging samples such as low-volume, low- concentration samples as they originate from photooxidation of PAH in the thin-film reactor.

The gas-phase dopant-delivery system has several key advantages, when compared to other dopant-delivery systems such as liquid delivery. The addition of the dopant in the nebulizer gas allows one to decouple the HPLC effluent from the dopant choice. Therefore any dopant

107 with sufficient vapor pressure can be used, regardless of immiscibility with the mobile phase composition. Additionally, the liquid HPLC flow path is kept free of unnecessary hardware, ensuring unhindered chromatographic separation. The dopant-delivery system is a passive device with excellent delivery characteristics, allowing for a constant stream of dopant. The high capacity of the developed system minimizes handling and maintenance time, enabling the system to be used in routine environments.

Dopant-assisted APPI-MS with benzene as dopant was established as a novel qualitative technique to assess the identities of OPAC. The use of benzene as dopant allows the broad ionization of many types of compounds, independent of their functionality. Therefore dopant- assisted APPI-MS with benzene as dopant can be utilized as an universal qualitative method. The establishment of rules for the ionization patterns of different compound classes was crucial for the identification of polycyclic aromatic compounds and can be applied to simple as well as complex cases.

6.2. Newly Identified UV Oxidation Products

By utilizing the analytical advances developed for the chemical analysis of OPAC, it was possible to identify several compounds for the first time as oxidation products of PAH in aqueous solutions. 4,5-pyrenequinone was identified for the first time as an oxidation product of the photodegradation of pyrene in the presence of only water and air. 2-hydroxy-9,10- phenanthrenequinone and 6H-dibenzo[b,d]pyran-6-ol were also identified for the first time as products of any type of chemical or photochemical degradation of phenanthrene and 9H- fluorene, respectively. The identification of these products allows one to confirm the addition of

108 molecular or singlet oxygen to the PAH as the main mechanism of PAH photodegradation in the presence of water and air.

6.3. Interfacial Degradation of 9H-Fluorene

In order to highlight the influence of the interface on the degradation, the photochemical degradation of 9H-fluorene was assessed in water films of two different thicknesses. Increasing the surface-to-volume ratio by a factor of five enhanced the degradation of 9H-fluorene significantly, leading to the increased rate of formation of the known photooxidation products,

9H-fluorene-9-one, 9H-fluorene-9-ol and 6H-dibenzo[b,d]pyran-6-ol. The kinetic evaluation of the formation of the oxidation products of 9H-fluorene allowed the establishment of a probable reaction pathway, with 9H-fluorene-9-one as stable product, and 9H-fluorene-9-ol and 6H- dibenzo[b,d]pyran-6-ol as reactive intermediates.

The enhanced reaction rate of 9H-fluorene at the interface is important for atmospheric studies, as it potentially reduces the lifetime of volatile PAH such as 9H-fluorene in the gas phase. Actual atmospheric hydrometeors with large surface-to-volume ratios could therefore transform non-polar PAH very quickly into their polar, oxygenated counterparts and withdraw the PAH in form of their oxidation products via uptake into the bulk liquid from the gas-phase.

6.4. Suggested Future Work

The developed application of dopant-assisted APPI-MS as a universal, qualitative analytical technique prompts the suggestion, to further utilize this technique for the identification of polycyclic aromatic compounds. Current developments in fuel technology, the shift to oxygen-containing fuels such as biodiesel will require the re-evaluation of the emission

109 characteristics of these fuels.172 The oxygen incorporated in these fuels might reduce the formation of pollutants such as polycyclic aromatic hydrocarbons, but it might also result in an increase of OPAC, a compound class which has been neglected in emission monitoring. Dopant- assisted APPI-MS would be ideal for the qualitative and quantitative assessment of these new types of emissions.

The analysis of PAH photodegradation products is also an area, in which there are ample opportunities for additional research waiting. In particular the analysis of the photodegradation products of PAH, such as acenaphthene, whose photodegradation has been neglected in the literature, promises a challenging and rewarding field.

The principle of interfacial degradation itself provides many possible avenues for future work: As the major photodegradation products of 9H-fluroene have been identified, it would be sensible to extend the kinetic analysis of 9H-fluorene’s photodegradation. Comparative studies, which investigate the effects of pH, salinity and inorganic ions, could be therefore based on the photodegradation of 9H-fluorene. For the simulation of real-world conditions, water collected from atmospheric hydrometeors such as rain, snow or fog, could provide additional insight and provide environmentally relevant conditions.

The observed enhanced photodegradation of PAH in thin-water films, with rate increases of more than 10-fold, suggests the use of the thin-film photooxidation for advanced technical applications. Photochemical reactions are usually fairly inefficient, as the light sources are expensive and require substantial amounts of energy. Conducting photochemical reactions, in which, instead of bulk phases, small droplets with large surface-to-volume ratios are irradiated, might therefore improve the efficiency of photochemical reactors significantly.

110 REFERENCES

1. 42 U.S.C. § 7401-7671q, 40 CFR 50-88

2. Beeson, W. L.; Abbey, D. E.; Knutsen, S. F. Environmental Health Perspectives 1998, 106, 813-822.

3. Gilmour, M. I.; Jaakkola, M. S.; London, S. J.; Nel, A. E.; Rogers, C. A. Environmental Health Perspectives 2006, 114, 627-634.

4. Department of Transportation, Bureau of Transportation Statistics; National Transportation Statistics 2008 accessed at http://www.bts.gov/publications/national_transportation_statistics/ on 07/10/08

5. NOAA, A costly phenomenen, accessed at http://www.noaa.gov/features/economic_1108/fog.html on 07/10/11

6. Calvert, J. G. Environmental Science & Technology 1976, 10, 256-262.

7. Salama, F.; Galazutdinov, G. A.; Krelowski, J.; Allamandola, L. J.; Musaev, F. A. Astrophysical Journal 1999, 265-273.

8. Allamandola, L. J.; Hudgins, D. M.; Sandford, S. A. Astrophysical Journal 1999, 511, L115-L119.

9. Hudgins, D. M.; Allamandola, L. J. Astrophysical Journal 1999, 516, L41-L44.

10. Ehrenfreund, P.; Rasmussen, S.; Cleaves, J.; Chen, L. Astrobiology 2006, 6, 490-520.

11. Platts, S. N., Rensselaer Polytechnic Institute, 2006.

12. United States Environmental Protection Agency; Health Assessment Document for Diesel Engine Exhaust. edited by National Center for Environmental Assessment and Office of Research and Development, 669. Washington, DC, 2002. EPA/600/8-90/057F

13. Poddar, N. B.; Thomas, S.; Wornat, M. J. Proceedings of the Combustion Institute 2011, 33, 541-548.

14. Poddar, N. B.; Thomas, S.; Ehrenhauser, F. S.; Wornat, M. J. Proceedings of the Combustion Institute 2009, 32, 1847-1854.

15. Wornat, M. J.; Ledesma, E. B.; Marsh, N. D. Fuel 2001, 80, 1711-1726.

16. Thomas, S.; Wornat, M. J. International Journal of Environmental Analytical Chemistry 2008, 88, 825-840.

17. 33 U.S.C. §1251 -1376 40 C.F.R. 104-149 section 303 (d)

111 18. United States Environmental Protection Agency; Appendix H - Estimating Carcinogenic Potency for Mixtures of Polycyclic Organic Matter (POM) for the 1996 National-Scale Assessment. in EPA-453/R-01-003

19. Tasdemir, Y.; Esen, F. Atmospheric Environment 2007, 41, 1288-1301.

20. Dabestani, R.; Ivanov, I. N. Photochemistry and Photobiology 1999, 70, 10-34.

21. Gómez-Gutiérrez, A.; Eric Jover, J. M. B.; Albaigés, J. Analytica Chimica Acta 2007, 583, 202-209.

22. Ma, Y.-G.; Lei, Y. D.; Xiao, H.; Wania, F.; Wang, W.-H. Journal of Chemical & Engineering Data 2010, 55, 819-825.

23. Lindsey, M. E.; Tarr, M. A. Water Research 2000, 34, 2385-2389.

24. Aubry, J. M.; Pierlot, C.; Rigaudy, J.; Schmidt, R. Accounts of Chemical Research 2003, 36, 668-675.

25. Ledakowicz, S.; Miller, J. S.; Olejnik, D. International Journal of Photoenergy 2001, 3, 95-101.

26. Atkinson, R. Chemical Reviews 1986, 86, 69-201.

27. Hewitt, C. N.; Harrison, R. M. Atmospheric Environment (1967) 1985, 19, 545-554.

28. Faust, B. C.; Allen, J. M. Environmental Science & Technology 1993, 27, 1221-1224.

29. Anastasio, C.; McGregor, K. G. Atmospheric Environment 2001, 35, 1079-1089.

30. Prinn, R. G. Annual Review of Environment and Resources 2003, 28, 29-57.

31. Gudipati, M. S.; Allamandola, L. J. The Astrophysical Journal 2004, 615, L177-L180.

32. Woon, D. E.; Park, J.-Y. The Astrophysical Journal 2004, 607, 342-345.

33. Ibrahim, Y. M.; Meot-Ner, M.; Alshraeh, E. H.; El-Shall, M. S.; Scheiner, S. Journal of the American Chemical Society 2005, 127, 7053-7064.

34. Pfleiderer, G.; Riedl, H. Process for preparing hydrogen peroxide. US2369912, 1945.

35. Carapellucci, P. A.; Wolf, H. P.; Weiss, K. Journal of the American Chemical Society 1969, 91, 4635-4639.

36. Debestani, R.; Ellis, K. J.; Sigman, M. E. Journal of Photochemistry and Photobiology A: Chemistry 1995, 86, 231-239.

37. Graber, E. R.; Rudich, Y. Atmospheric Chemistry and Physics 2006, 6, 729-753.

112 38. Mmereki, B. T.; Donaldson, D. J.; Gilman, J. B.; Eliason, T. L.; Vaida, V. Atmospheric Environment 2004, 38, 6091-6103.

39. McConkey, B. J.; Hewitt, L. M.; Dixon, D. G.; Greenberg, B. M. Water, Air, and Soil Pollution 2002, 136, 347-359.

40. Punjabi, P. B.; Ameta, R.; Kumar, A.; Jain, M. Bulletin of the Chemical Society of Ethiopia 2008, 22, 361-368.

41. Sigman, M. E.; Zingg, S. P.; Pagni, R. M.; Burns, J. H. Tetrahedron Letters 1991, 32, 5737-5740.

42. Sigman, M. E.; Schuler, P. F.; Ghosh, M. M.; Dabestani, R. T. Environmental Science & Technology 1998, 32, 3980-3985.

43. Miller, J. S.; Olejnik, D. Water Research 2001, 35, 233-243.

44. Trapido, M.; Veressinina, Y.; Munter, R. Environmental Technology 1995, 16, 233-241.

45. Tham, Y.; Sakugawa, H. Bulletin of Environmental Contamination and Toxicology 2007, 79, 670-673.

46. Fasnacht, M. P.; Blough, N. V. Aquatic Sciences - Research Across Boundaries 2003, 65, 352-358.

47. Fasnacht, M. P.; Blough, N. V. Environmental Science & Technology 2003, 37, 5767- 5772.

48. Fasnacht, M. P.; Blough, N. V. Environmental Science & Technology 2002, 36, 4364- 4369.

49. Sabaté, J.; Bayona, J. M.; Solanas, A. M. Chemosphere 2001, 44, 119-124.

50. Fox, M. A.; Olive, S. Science 1979, 205, 582-583.

51. Katz, M.; Chan, C.; Tosine, H.; Sakuma, T. In Polynuclear Aromatic Hydrocarbons; Jones, P. W.; Leber, P. Eds.; Ann Arbor Science Publishers: Ann Arbor, MI, 1979; pp. 171-189.

52. Gerasimov, G. Y. High Energy Chemistry 2004, 38, 161-166.

53. Nielsen, T.; Ramdahl, T.; Bjørseth, A. Environmental Health Perspectives 1983, 47.

54. Korfmacher, W. A.; Natusch, D. F. S.; Taylor, D. R.; Mamantov, G.; Wehry, E. L. Science 1980, 207, 763-765.

55. Beitrán, F. J.; González, M.; Rivas, F. J. Journal of Environmental Science and Health . Part A: Environmental Science and Engineering and Toxicology 1996, 31, 2193-2210.

113 56. Bertilsson, S.; Widenfalk, A. Hydrobiologia 2002, 469, 23-32.

57. Sigman, M. E.; Chevis, E. A.; Brown, A.; Barbas, J. T.; Dabestani, R.; Burch, E. L. Journal of Photochemistry and Photobiology A: Chemistry 1996, 94, 149-155.

58. Mallakin, A.; Dixon, D. G.; Greenberg, B. M. Chemosphere 2000, 40, 1435-1441.

59. Trapido, M.; Veressinina, Y.; Munter, R. Environmental Technology 1995, 16, 729-740.

60. Swallow, K. C.; Taghizadeh, K.; Weakland, S.; Plummer, E. F.; Busby, W. F.; Lafleur, A. L. International Journal of Environmental Analytical Chemistry 1995, 60, 113-122.

61. Reyes, C.; Sigman, M. E.; Arc, R.; Barbas, J. T.; Dabestani, R. Journal of Photochemistry and Photobiology A: Chemistry 1998, 112, 7.

62. Moeini-Nombel, L.; Matsuzawa, S. Journal of Photochemistry and Photobiology A: Chemistry 1998, 119, 15-23.

63. Milano, J. C.; Tallone-Maesano, E.; Vernet, J.-L. Environmental Technology 1999, 20, 1019-1032.

64. Mallakin, A.; McConkey, B. J.; Miao, G.; McKibben, B.; Snieckus, V.; Dixon, D. G.; Greenberg, B. M. Ecotoxicology and Environmental Safety 1999, 43, 204-212.

65. Matsuzawa, S. Polycyclic Aromatic Compounds 2000, 21, 331-339.

66. Patel, J. R.; Politzer, I. R.; Griffin, G. W.; Laseter, J. L. Biological Mass Spectrometry 1978, 5, 664-670.

67. Romeas, V.; Pichat, P.; Guillard, C.; Chopin, T.; Lehaut, C. Industrial & Engineering Chemistry Research 1999, 38, 3878-3885.

68. R. Świetlik, D. K., J. Dojlido Polish Journal of Environmental Studies 2002, 11, 165-169.

69. Grannas, A. M.; Bausch, A. R.; Mahanna, K. M. The Journal of Physical Chemistry A 2007, 111, 11043-11049.

70. Kahan, T. F.; Donaldson, D. J. The Journal of Physical Chemistry A 2007, 111, 1277- 1285.

71. Chen, J.; Ehrenhauser, F. S.; Valsaraj, K. T.; Wornat, M. J. The Journal of Physical Chemistry A 2006, 110, 9161-9168.

72. Chen, J.; Valsaraj, K. T. The Journal of Physical Chemistry A 2007, 111, 4289-4296.

73. Chen, J.; Ehrenhauser, F. S.; Valsaraj, K. T.; Wornat, M. J. In Atmospheric Aerosols Characterization, Chemistry, Modeling, and Climate; Kalliat T. Valsaraj, R. R. K. Ed.; Oxford University Press: Washington, DC, 2009; pp. 127-146.

114 74. Ghosal, S.; C.Hemminger, J.; Bluhm, H.; Mun, B. S.; Hebenstreit, E.; Ketteler, G.; Ogletree, D. F.; Requejo, F. G.; Salmeron, M. Science 2005, 307, 563-566.

75. Jungwirth, P.; Winter, B. The Annual Review of Physical Chemistry 2008, 59, 343-366.

76. Clifford, D.; Donaldson, D. J. The Journal of Physical Chemistry A 2007, 111, 9809- 9814.

77. Herckes, P.; Hannigan, M. P.; Trenary, L.; Lee, T.; Collett, J. L. Atmospheric Research 2002, 64, 99-108.

78. Capel, P. D.; Leuenberger, C.; Giger, W. Atmospheric Environment. Part A. General Topics 1991, 25, 1335-1346.

79. Valsaraj, K. T.; Thoma, G. J.; Reible, D. D.; Thibodeaux, L. J. Atmospheric Environment 1993, 27A, 203-210.

80. Chen, J. Adsorption and Photochemical Transformations of Polycyclic Aromatic Hydrocarbons on Atmospheric Water Films. Louisiana State University, Baton Rouge, 2008.

81. Vácha, R.; Jungwirth, P.; Chen, J.; Valsaraj, K. Physical Chemistry Chemical Physics 2006, 8, 4461-4467.

82. Raymond, E. A.; Tarbuck, T. L.; Brown, M. G.; Richmond, G. L. The Journal of Physical Chemistry B 2003, 107, 546-556.

83. Mmereki, B. T.; Donaldson, D. J. The Journal of Physical Chemistry A 2003, 107, 11038-11042.

84. Kahan, T. F.; Donaldson, D. J. Environmental Research Letters 2008, 3, 1-6.

85. Kahan, T. F.; Donaldson, D. J. Environmental Science & Technology 2010, 44, 3819- 3824.

86. Jacob, J. Pure and Applied Chemistry 1996, 68, 301-308.

87. Jacob, J. Polycyclic Aromatic Compounds 2008, 28, 242-272.

88. Yu, H. Journal of Environmental Science and Health Part C-Environmental Carcinogenesis & Ecotoxicology Reviews 2002, C20, 149-183.

89. Lewis, R. J. Naphthalene 91-20-3. In Sax's Dangerous Properties of Industrial Materials, John Wiley & Sons, Inc.: 2004.

90. Lewis, R. J. 1,4-Naphthoquinone 130-15-4. In Sax's Dangerous Properties of Industrial Materials, John Wiley & Sons, Inc.: 2004.

115 91. Lewis, R. J. 1,2-Naphthoquinone 524-42-5. In Sax's Dangerous Properties of Industrial Materials, John Wiley & Sons, Inc.: 2004.

92. D. J. Donaldson, K. T. V. Environmental Science & Technology 2010, 44, 865-873.

93. Chang, D. P. Y.; Hill, R. C. Atmospheric Environment 1980, 14, 803-807.

94. Russel, L. M.; Noone, K. J.; Ferek, R. J.; Pckalny, R. A.; Flagan, R. C.; Seinfeld, J. H. Journal of the Atmospheric Sciences 2000, 57, 2591-2606.

95. Atkinson, R.; Lloyd, A. C. Evaluation of Kinetic and Mechanistic Data for Modeling of Photochemical Smog; NIST, 1984; Vol. 13.

96. Norikazu Nagae; Enami, T.; Doshi, S. LCGC North America 2002, 20, 964-972.

97. Syage, J. A.; Hanold, K. A.; Lynn, T. C.; Julie A. Horner; Thakur, R. A. Journal of Chromatography A 2004, 1050, 137-149.

98. Robb, D. B.; Covey, T. R.; Bruins, A. P. Analytical Chemistry 2000, 72, 3653-3659.

99. Syage, J. A.; Evans, M. D.; Hanold, K. A. American Laboratory 2000.

100. Short, L. C.; Cai, S.-S.; Syage, J. A. Journal of the American Society for Mass Spectrometry 2007, 18, 589-599.

101. Bos, S. J.; Leeuwen, S. M. v.; Karst, U. Analytical and Bioanalytical Chemistry 2006, 384, 85-99.

102. Raffaelli, A.; Saba, A. Mass Spectrometry Reviews 2003, 22, 318-331.

103. Marchi, I.; Rudaz, S.; Veuthey, J.-L. Talanta 2009, 78, 1-18.

104. Robb, D. B.; Blades, M. W. Analytica Chimica Acta 2008, 627, 34-49.

105. Walker, M. S.; Wornat, M. J. Journal of Chromatography A 2010, 1217, 4568-4574.

106. McClaine, J. W.; Zhang, X.; Wornat, M. J. Journal of Chromatography A 2006, 1127, 137-146.

107. Müller, A.; Mickel, M.; Geyer, R.; Ringseis, R.; Eder, K.; Steinhart, H. Journal of Chromatography B 2006, 837, 147-152.

108. Hanold, K. A.; Fischer, S. M.; Cormia, P. H.; Miller, C. E.; Syage, J. A. Analytical Chemistry 2004, 76, 2842-2852.

109. Kauppila, T. J.; Kotiaho, T.; Bruins, A. P.; Kostiainen, R. Journal of the American Society for Mass Spectrometry 2004, 15, 203-211.

116 110. Traldi, P.; Marotta, E. In Advances in Mass Spectrometry; Ashcroft, A. E.; Brenton, G.; Monaghan, J. J. Eds., 2004; pp. 275-293.

111. Kauppila, T. J.; Kostiainen, R.; Bruins, A. P. Rapid Communications in Mass Spectrometry 2004, 18, 808-815.

112. Marotta, E.; Seraglia, R.; Fabris, F.; Traldi, P. International Journal of Mass Spectrometry 2003, 228, 841-849.

113. Tubaro, M.; Marotta, E.; Seraglia, R.; Traldi, P. Rapid Communications in Mass Spectrometry 2003, 17, 2423-2429.

114. Robb, D. B.; Blades, M. W. Journal of the American Society for Mass Spectrometry 2005, 16, 1275.

115. Robb, D. B.; Blades, M. W. Analytical Chemistry 2006, 78, 8162-8164.

116. Kauppila, T. J.; Kuuranne, T.; Meurer, E. C.; Eberlin, M. N.; Kotiaho, T.; Kostiainen, R. Analytical Chemistry 2002, 74, 5470-5479.

117. Hanold, K. A.; Evans, M. D.; Syage, J. A.; Fischer, S. M.; Cormia, P. H., Photoionization: Direct APPI vs. Dopant APPI. ASMS Meeting, Chicago, IL, USA 2001.

118. Cai, S.-S.; Syage, J. A.; Karl A. Hanold; Balogh, M. P. Analytical Chemistry 2009, 81, 2123-2128.

119. Hollosi, L.; Wenzl, T. Journal of Chromatography A 2011, 1218, 23-31.

120. Syage, J. A. Journal of the American Society for Mass Spectrometry 2004, 15, 1521- 1533.

121. Cai, Y.; Kingery, D.; McConnell, O.; Bach, A. C. Rapid Communications in Mass Spectrometry 2005, 19, 1717-1724.

122. Grosse, S.; Letzel, T. Journal of Chromatography A 2007, 1139, 75-83.

123. Itoh, N.; Narukawa, T.; Numata, M.; Aoyagi, Y.; Yarita, T.; Takatsu, A. Polycyclic Aromatic Compounds 2009, 29, 41-55.

124. Himmelsbach, M.; Buchberger, W.; Reingruber, E. Polymer Degradation and Stability 2009, 94, 1213-1219.

125. Zheng, J.; Shamsi, S. A. Analytical Chemistry 2006, 78, 6921-6927.

126. Itoh, N.; Aoyagi, Y.; Yarita, T. Journal of Chromatography A 2006, 1131, 285-288.

127. Himmelsbach, M.; Haunschmidt, M.; Buchberger, W.; Klampfl, C. W. Journal of Chromatography A 2007, 1159, 58-62.

117 128. Kauppila, T. J.; Kostiainen, R.; Bruins, A. P. Journal of the American Society for Mass Spectrometry 2005, 16, 1399-1407.

129. Linstrom, P. J.; Mallard, W. G., Eds. NIST Chemistry WebBook; National Institute of Standards and Technology: Gaithersburg MD.

130. Ehrenhauser, F. S.; Wornat, M. J.; Valsaraj, K. T.; Rodriguez, P. Rapid Communications in Mass Spectrometry 2010, 24, 1351-1357.

131. Stein, S.; Mirkohin, Y. Tchekhovskoi, D.; Mallard, G. The NIST Mass Spectral Search Program for the NIST/EPA/NIH Mass Spectral Library,Version 2.0; 2004

132. Oña, J. O.; Wornat, M. J. Polycyclic Aromatic Compounds 2008, 28, 15 - 38.

133. Oña, J. O.; Wornat, M. J. Polycyclic Aromatic Compounds 2007, 27, 165 - 183.

134. Moriwaki, H. The Analyst 2000, 125, 417-420.

135. Takinoa, M.; Daishima, S.; Yamaguchi, K.; Nakahara, T. Journal of Chromatography A 2001, 928, 53-61.

136. Marvina, C. H.; Smith, R. W.; Bryant, D. W.; McCarry, B. E. Journal of Chromatography A 1999, 863, 13-24.

137. Lafleur, A. L.; Taghizadeh, K.; Howard, J. B.; Anacleto, J. F.; Quilliam, M. A. Journal of the American Society for Mass Spectrometry 1996, 7, 276-286.

138. Bordwell, F. G. Accounts of Chemical Research 1988, 21, 456-463.

139. Tubaro, M.; Marotta, E.; Seraglia, R.; Traldi, P. Rapid Communications in Mass Spectrometry 2003, 17, 2423-2429.

140. Bartmess, J. E.; Scott, J. A.; Robert T. McIver, J. Journal of the American Chemical Society 1979, 101, 6046-6056.

141. Stein, S. E.; Roumiantsev, A. K. ISOFORM v1.2 - NIST Isotope Calculator, 1997

142. Raja, S.; Valsaraj, K. T. Environmental Science & Technology 2003, 38, 763-768.

143. Davidovits, P.; Kolb, C. E.; Williams, L. R.; Jayne, J. T.; Worsnop, D. R. Chemical Reviews 2006, 106, 1323-1354.

144. Klöpffer, W.; Haag, F.; Kohl, E.-G.; Frank, R. Ecotoxicology and Environmental Safety 1988, 15, 298-319.

145. Ali, L. N.; Mantourab, R. F.; Rowland, S. J. Marine Environmental Research 1995, 40, 319-335.

118 146. Dowty, B. J.; Brightwell, N. E.; Laseter, J. L.; Griffin, G. W. Biochemical and Biophysical Research Communication 1974, 57, 452-456.

147. Newman, M. S.; Childers, R. L. The Journal of Organic Chemistry 1967, 32, 62-66.

148. Werner, A.; Löwenstein, B.; Wack, A.; Frey, T.; Kunz, M.; Rekner, K.; Ney, A.; Heil, H.; Scherrer, A.; Schwabacher, H.; Kunz, J.; Grob, A. Justus Liebigs Annalen der Chemie 1902, 321, 248-357.

149. Werner, A.; Löwenstein, B.; Wack, A.; Frey, T.; Kunz, M.; Rekner, K.; Ney, A.; Heil, H.; Scherrer, A.; Schwabacher, H.; Kunz, J.; Grob, A. Justus Liebigs Annalen der Chemie 1902, 322, 135-173.

150. Lee, J.; Lane, D. A. Atmospheric Environment 2010, 44, 2469-2477.

151. Miet, K.; Budzinski, H.; Villenave, E. Polycyclic Aromatic Compounds 2009, 29, 267 - 281.

152. Krishnan, S.; Kuhn, D. G.; Hamilton, G. A. Journal of the American Chemical Society 1977, 99, 8122-8124.

153. Chen, P. N.; Junk, G. A.; Svec, H. J. Environmental Science & Technology 1979, 13, 451-4.

154. Reyes, C.; Sigman, M. E.; Arce, R.; Barbas, J. T.; Dabestani, R. Journal of Photochemistry and Photobiology A: Chemistry 1998, 112, 277-283.

155. Johannes, C.; Majcherczyk, A.; Hüttermann, A. Journal of Biotechnology 1998, 61, 151- 156.

156. Callighan, R. H.; Tarker, M. F., Jr.; Wilt, M. H. Journal of Organic Chemistry 1962, 27, 765-8.

157. Shemer, H.; Linden, K. G. Water Research 2007, 41, 853-859.

158. Banzatti, C.; Salvadori, P.; Tonani, R. Indian Journal of Chemistry 1984, 23B, 671-672.

159. Sollazzo, G.; Tonani, R.; Traldi, P. Organic Mass Spectrometry 1988, 23, 550-552.

160. Moliner, J. R.; Coy, F. P.; Llorente, L. C.; Banus, M. C.; Riudavets, J. A. C.; Rofes, C.; Rofes, N.; Siso, A. F.; Marcas, O. d. E. d. P. y. Ed.: Spain, 1997.

161. Banzatti, C.; Branzoli, U.; Lovisolo, P. P.; Melloni, P.; Orsini, G.; Salvadori, P. Arzneimittel-Forschung 1984, 34, 864-869.

162. Dewar, P. S.; Forrester, A. R.; Thomson, R. H. Journal of the Chemical Society D: Chemical Communications 1970, 850-850.

119 163. Dewar, P. S.; Forrester, A. R.; Thomson, R. H. Journal of the Chemical Society C: Organic 1971, 3950-3959.

164. Mehta, G.; Pandey, P. N. Synthesis 1975, 404-5.

165. Liyana-Arachchi, T. P.; Valsaraj, K. T.; Hung, F. R. The Journal of Physical Chemistry A 2011, 115, 9226-9236.

166. Teschke, O.; Souza, E. F. d. Chemical Physical Letters 2005, 403, 95-101.

167. Valsaraj, K. T. Environmental Toxicology and Chemistry 2004, 23, 2318-2323.

168. Fogler, H. S. Elements of Chemical Reaction Engineering, 3rd ed.; Prentice Hall New Jersey, 2003.

169. Barbas, J. T.; Sigman, M. E.; Arce, R.; Dabestani, R. Journal of Photochemistry and Photobiology A: Chemistry 1997, 109, 229-236.

170. Luo, Y.-R. Handbook of bond dissociation energies in organic compounds; CRC Press: Boca Raton, 2002.

171. Renz, M.; Meunier, B. European Journal of Organic Chemistry 1999, 1999, 737-750.

172. Swanson, K. J.; Madden, M. C.; Ghio, A. J. Environmental Health Perspectives 2007, 115, 4.

173. Arakawa, T.; Sato, Y.; Ueno, T.; Funatsu, T.; Shoji, S. ―Pinhole-free pyrex glass etching th using HF-H2SO4 mixed acid and its application for APDMS microflow system‖; The 13 International Conference on Solid-State Sensors, Actuators and Miocrosystems, 2005, Seoul Korea.

174. Eisenberg, W. C.; Taylor, K.; Murray, R. W. Carcinogenesis 1984, 5, 1095-1096.

175. Swalheim, D. A.; Stapf, C. J., Jr. Preprint Papers - American Chemical Society, Division of Fuel Chemistry 1960, 24-33.

176. Cohen, F. L.; Cormier, U. Journal of the American Chemical Society 1930, 52, 4363- 4364.

177. Fatiadi, A. J. Journal of Chromatography 1965, 20, 319-324.

178. Ruhland, K.; Brück, A.; Herdtweck, E. European Journal of Inorganic Chemistry 2007, 2007, 944-964.

179. National Institute of Advanced Industrial Science and Technology (AIST); Spectral Database for Organic Compounds SDBS; acessed at http://riodb01.ibase.aist.go.jp/sdbs/cgi-bin/cre_index.cgi 08/01/2011

180. Fieser, L. F. Organic Syntheses 1936, 16, 63.

120 181. Gattermann, L.; Wieland, H. Laboratory Methods of Organic Chemistry; The MacMillan Company: New York, USA, 1937.

182. Alarcón, S. H.; Olivieri, A. C.; Labadie, G. R.; Cravero, R. M.; González-Sierra, M. Tetrahedron 1995, 51, 4619-4626.

121 APPENDIX A: HPLC METHODS

Method A

Method name: VYDACA.M

Instrument: Hewlett Packard 1050 HPLC system with quaternary pump (79852A), manual 6-port

injection, column compartment (79852) and UV-DAD detector (G1301A)

Column: 4.6×250 mm 5 µm Vydac 201TP54 (W. R. Grace & Co.-Conn., Columbia, MD, USA)

Injection volume: 20 μL

Column temperature: 30°C

Flow rate: 1.5 mL/min

Solvent program:

Start at 40/60% acetonitrile/water, ramp to 100% acetonitrile within 40 min, ramp to

100% dichloromethane within 40 min, ramp to 100% acetonitrile within 10 min, ramp to

40/60% acetonitrile/water within 10 min an hold at 40/60% acetonitrile/water for 10 min.

UV/DAD settings:

Wavelength range: 190-520 nm without reference

Spectral resolution: 2 nm

Slit: 4 nm

Sampling frequency: every 1.3 s

122 Method B

Method name: FSEN50A.M

Instrument: Agilent 1100 system with quaternary pump (G1311A) with vacuum degasser

(G1379), autosampler (G1313A), column compartment (G1316), UV-DAD detector

(G1315B) and APPI-MS (G1956B).

Column: 2.1×250 mm 5 µm Pinnacle II Phenyl column (Restek Corp., Bellefonte, PA, USA)

Injection volume: 50 μL

Column temperature: 30°C

Flow rate: 0.2 mL/min

Solvent program:

Start at 50/50% acetonitrile/water and hold for 30 min, ramp to 100% acetonitrile within

30 min and for 30 min, ramp to 50/50% acetonitrile/water within 10 min.

UV/DAD settings:

Wavelength range: 252-256 nm without reference

Spectral resolution: 2 nm

Slit: 4 nm

Sampling frequency: every 2s

APPI-MS settings:

Spray Chamber: 9 L/min drying gas flow rate; +1400 V/-1800 V capillary voltage; 35 psig

nebulizer pressure; 350°C drying gas temperature; 350°C vaporizer

temperature.

MS Signal Settings: 1 positive / 1 negative signal; 100-400 m/z range; 70 V fragmentor voltage;

1.0 gain; 150 cps threshold; 0.25 step size; 0.97 sec/cycle.

123 Method C

Method name: FSEM00B1.M

Instrument: Agilent 1100 system with quaternary pump (G1311A) with vacuum degasser

(G1379), autosampler (G1313A), column compartment (G1316), UV-DAD detector

(G1315B) and APPI-MS (G1956B).

Column: 2.1×250 mm 5 µm Ultra Aqueous C18 Column (Restek Corp., Bellefonte, PA, USA)

Injection volume: 50 μL

Column temperature: 40°C

Flow rate: 0.2 mL/min

Solvent program:

Start at 100% water, hold at 100% water for 5 min, ramp to 100% methanol within 30

min, hold at 100% methanol for 30 min, ramp to 100% water within 10 min.

UV/DAD settings:

Wavelength range: 252-256 nm without reference

Spectral resolution: 2 nm

Slit: 4 nm

Sampling frequency: every 0.2 s

APPI-MS settings:

Spray Chamber: 9 L/min drying gas flow rate; +/-1800 V capillary voltage; 35 psig nebulizer

pressure; 350°C drying gas temperature; 350°C vaporizer temperature.

MS Signal Settings: 1 positive / 1 negative signal; 2-500 m/z range; variable fragmentor voltage;

1.0 gain; 0 cps threshold; 0.25 step size; 0.95 sec/cycle.

124 Method D

Method name: FSEFOG3.M

Instrument: Agilent 1100 system with quaternary pump (G1311A) with vacuum degasser

(G1379), autosampler (G1313A), column compartment (G1316), UV-DAD detector

(G1315B) and APPI-MS (G1956B).

Column: 2.1×150 mm 3 µm Ultra Aqueous C18 Column (Restek Corp., Bellefonte, PA, USA)

Injection volume: 50 μL

Column temperature: 40°C

Flow rate: 0.2 mL/min

Solvent program:

Start at 100% water, hold at 100% water for 3 min, ramp to 100% methanol within 30

min, hold at 100% methanol for 30 min, ramp to 100% water within 20 min.

UV/DAD settings:

Wavelength range: 252-256 nm without reference

Spectral resolution: 2 nm

Slit: 4 nm

Sampling frequency: every 0.5 s

APPI-MS settings:

Spray Chamber: 9 L/min drying gas flow rate; +/-1800 V capillary voltage; 35 psig nebulizer

pressure; 350°C drying gas temperature; 350°C vaporizer temperature.

MS Signal Settings: 1 positive / 1 negative signal; 90-500 m/z range; 150 V fragmentor voltage;

1.0 gain; 100 cps threshold; 0.25 step size; 0.96 sec/cycle

125 Method E

Method name: FSEFLUO.M

Instrument: Hewlett Packard 1100 HPLC system with quaternary ump (G1311A) with vacuum

degasser (G1322A), autosampler (G1313A), column compartment, UV-DAD detector

(G1315A) and HP1046A programmable fluorescence detector.

Column: 2.1×250 mm 4 µm Pinnacle II PAH Column (Restek Corp., Bellefonte, PA, USA)

Injection volume: 4 μL

Column temperature: 35°C

Flow rate: 0.2 mL/min

Solvent program: 15 min isocratic at 80% acetonitrile / 20% water

UV/DAD settings:

Wavelength range: 252-256 nm without reference

Spectral resolution: 2 nm

Slit: 4 nm

Sampling frequency: every 2 s

FLD settings:

Excitation wavelength: 225 nm

Emission wavelength: 315 nm

PMTgain: 13

Lamp frequency: 55 Hz

Responsetime: 0.25 s

126 Method F

Method name: FEFTM3.M

Instrument: Hewlett Packard 1100 HPLC system: quaternary ump (G1311A) with vacuum

degasser (G1322A), autosampler (G1313A), column compartment, UV-DAD detector

(G1315A) and HP1046A programmable fluorescence detector (not used).

Column: 2.1×50 mm 5 µm Polar-RP Column (Phenomenex, Torrance, CA, USA)

Injection volume: 4 μL

Column temperature: 15°C

Flow rate: 0.2 mL/min

Solvent program:

Start at 80/20% water/acetonitrile, hold at 80/20% water/acetonitrile for 0.1 min, ramp to

100% acetonitrile within 6.9 min, hold at 100% acetonitrile for 4 min, ramp to 80/20%

water/acetonitrile within 2 min.

UV/DAD settings:

Wavelength range: 252-256 nm

Reference: none used

Spectral resolution: 2 nm

Slit: 4 nm

Sampling frequency: every 2 s

127 APPENDIX B: GLASS SLIDE PREPARATION

The core piece of the thin-film reactor is the glass slide on which the thin water film rests.

The in previous work used open-ended, rectangular U-profile, obtained by cutting a rectangular borosilicate pipe,80 was modified to enhance handling and the stability of the thin film stability.

The improved solution was accomplished by etching a rectangular recession into a

40 mm×1000 mm glass plate with 2 mm thickness (Technical Glass Products Inc., Painesville

Township, OH, USA), creating an effective trough in which the liquid is well contained on all sides and uneven liquid distribution due to adhesion to the side-walls is limited. Borosilicate glass is one of the most chemically inert glasses, hence its preferred use for chemical glassware and in this reactor. The advantageous property of its chemical inertness makes borosilicate glass also immune to etching with common commercially available glass etching cream. As borosilicate glass is only second to pure quartz glass in its inertness, its etching can only be accomplished by using hydrogen fluoride (HF). HF-etching is very common in the semiconductor industry, but unfortunately the corresponding etching procedures are usually employed on a fairly small scale (450 mm as maximum wafer diameter) and the usual HF gas etching chambers cannot accommodate the long thin-film-reactor glass slide. Therefore a wet etching process, utilizing hydrofluoric acid, was deployed. Hydrofluoric acid is a very strong agent, commonly causing etching mistakes like pinholes and roughness. To allow for a smooth and even glass removal, hydrofluoric acid was deployed as dilute solution in sulfuric acid following the work of Arakawa et. al.173 Sodium fluoride can substitute for the hydrofluoric acid as it converts in the presence of sulfuric acid completely to hydrofluoric acid and it is easier and safer to handle. Initial tests showed that the high concentration of sodium ions caused a

128 precipitation layer to form (most likely Na2SO4·10 H2O), which inhibited the etching process significantly, causing unreasonably long etching times.

Apart from the etching reagent, a suitable mask is also required to allow for the desired pattern to be fabricated. For this work the mask had to be simple and effective, and its deployment had to be achievable without complicated equipment. Masks commonly used in the semiconductor industry, such as acid resistant photoresist or other high-tech coatings, (e.g. metal vapor deposition) are impossible to be deployed for the utilized glass plate because of its large dimensions (40 mm×900 mm). Therefore less sophisticated but readily available materials were tested. Table 10 gives an overview of the tested materials and some comments on their performance.

Table 10. Mask material for HF etching

Mask material Type of Coating Performance

3M Scotch Magic Tape Polyethylene terephthalate Partially degraded within PET one hour lifts during etching

Parafilm® M Polyolefine/paraffin wax No degradation, lifts very quickly

Nail Polish ―Sally Hansen Celluloseacetate No degradation, lifts during HARD as NAIL Xtreme etching Wear‖

Fast Dry Spray Acrylic polymer, latex No degradation, very little ColorPlace® lifting, flaking at the etches after 2-3 days

As can be seen in Table 10, acrylic spray paint was the most successful mask material, as it was not chemically attacked by the sulfuric acid/hydrofluoric acid mixture and avoided any significant lifting. The mask was created by using sticky tape such as 3M Magic Tape as template for the trough and spray-painting the rest of the slide. Careful removal of the tape and

129 the adhering paint left the trough without mask and made it therefore accessible to the acid for etching.

The etching process itself can be accomplished in two ways, either by adding the solution on the top of the glass slide, or by submerging the entire slide into an acid bath. The latter is the preferred method, as it allowed for continuous etching, whereas the former method needed replacement of the etching solution after approximately 30 min (after which a 0.8% HF/

30% H2SO4 solution is depleted of HF) and caused unnecessary mechanical stress on the mask due to the washing.

The submersion into an etch bath allowed for the etching of roughly 80 μm trough depth within three to four days. Slight lifting of the mask was only a minor problem, as diffusion limited undesired etching below the mask. The strength of the etching bath declined over time as some of the hydrogen fluoride evaporated. By adding fresh hydrofluoric acid every 12 hours, an etching rate of approximately 1-2 μm per hour could be accomplished.

After the etching process, the slide possessed a very hydrophilic surface. This surface caused a drop of water to spread out very effectively, overlapping even the etched boundary.

Therefore the glass slide had to be subjected to surface treatment to enable good containment of the liquid. As a first step, the entire glass slide surface was silanized by applying a solution of

10% (v/v) dichlorodimethylsilane in toluene twice onto the dry glass slide. After washing with methanol, which deactivated any residual chlorosilane groups, the entire slide exhibited a very hydrophobic surface, causing drops of any aqueous solutions to roll off without leaving any residue. The hydrophobic interior surface of the trough, which has to contain the water, was made wettable by removing the silanes through an alkaline etch process. For this purpose a 20% aqueous sodium hydroxide solution was distributed as multiple small drops inside the trough,

130 preferentially placed very close to each other and close to the trough edge, but without touching the edge or each other. The etching had to be done in this seemingly complicated manner, as otherwise the aqueous sodium hydroxide solution would have just rolled off the hydrophobic plate. After two hours, additional sodium hydroxide was applied, with the previously placed drops serving as anchor points for the additional solution. The addition of sodium hydroxide solution was continued until the entire trough was covered with liquid. After approximately twelve hours, when most of the sodium hydroxide had transformed into sodium (hydrogen) carbonate, it was neutralized with 10% acetic acid and washed off with deionized water, until the run-off was of neutral pH value. The finalized slide possessed a perfect hydrophilic trough with a hydrophobic boundary, containing aqueous solutions very well.

During the course of this work three slides have been prepared. Two slides with trough areas of 270.7 cm2 and 254.5 cm2 and trough depths of 274 µm and 63 µm, respectively, and an additional slide with three compartments (141.5 cm2, 112 µm; 94.1 cm2, 89 µm; 43.4 cm2, 112

µm) etched into it. The slide with 270.7 cm2 was utilized for the uptake and photolysis experiments, as it had the largest trough depth of 274 µm.

131 APPENDIX C: PURIFICATION OF POLYCYCLIC AROMATIC HYDROCARBONS

The purity of the utilized PAH is essential for their use in photochemical reactions. The most commonly encountered impurities in commercially available PAH are alkyl derivatives and other PAH. As the delivery of the PAH is accomplished through saturation in a packed-bed column, enrichment of more volatile impurities can occur. Furthermore, as all PAH can act as a photosensitizer (photocatalyst),20,174 even minor impurities might affect degradation rates prompting the essential need for the purification of commercial available (impure) PAH.

Naphthalene

Commercially available naphthalene (99+%, scintillation grade, Sigma Aldrich, St.

Louis, MO, USA) showed a high degree of purity with only a small amount of benzo[b]thiophene as impurity and was used without further purification. A potential cleaning procedure that utilizes sodium for binding sulfur-containing compounds was not undertaken due to the inherent danger of dispersed sodium and the already sufficient high purity of the naphthalene.175

Phenanthrene

Commercial phenanthrene (97%+, Acros Organics, Morris Plains, NJ, USA) contained anthracene, 9H-fluorene and other unknown compounds. Especially anthracene is an important impurity as it acts as a very effective photosensitizer. 9H-Fluorene is also an important compound to be removed, as its volatility is higher than that of phenanthrene, causing

132 enrichment in the case of gas-phase delivery. Other minor impurities were alkylated phenanthrenes, which are also undesirable, as they would lead to significantly different products

(oxidation of the methyl group instead of the aromatic body).

Suggested purification procedures in the literature, such as the recrystallization from ethanol with simultaneous oxidation with nitric acid176 or distillation over sodium metal, did not yield the desired purity. In both cases, new impurities from side reactions were found, and 9H- fluorene was not completely removed. Therefore a new purification procedure was designed, specifically targeted to remove 9H-fluorene and anthracene from phenanthrene.

The hydrogen at the 9-position of 9H-fluorene possesses significant acidity, which can be utilized to enhance the reactivity at this position. The acidity of 9H-fluorene is strong enough to allow the formation of a carbocation in dimethylsulfoxide (DMSO) with potassium hydroxide as a base. Therefore a preferred oxidation of 9H-fluorene to 9H-fluorene-9-one can be accomplished with hydrogen peroxide in DMSO in the presence of potassium hydroxide. The polar reaction product, 9H-fluorene-9-one, is easily separated from phenanthrene due to its enhanced solubility in a water/DMSO mixture, compared to the PAH.

6.17 g phenanthrene and 1 g potassium hydroxide were dissolved in 50 mL DMSO and stirred for 40 min. Two 2-mL aliquots of 30% hydrogen peroxide were added and the mixture stirred for a total for 2 hours, after which oxygen development had completely ceased. Upon addition of 100 mL of deionized water, the 9H-fluorene-free (<1 ppm) phenanthrene precipitated and was recovered via filtration.

Anthracene removal

The DMSO-wet phenanthrene was subsequently dissolved in 31.4 g ethanol, and 1.1 g concentrated nitric acid was added. After boiling at reflux for 8 minutes, water was added to

133 initiate crystallization. The resulting crystals were obtained by filtration and recrystallized from ethanol to yield 5.82 g anthracene-free (<1 ppm) and 9H-fluorene-free (<1 ppm) phenanthrene.

Alkylphenanthrene removal

The purified phenanthrene still contained alkylated compounds and was therefore subjected to photobromination in methanol, whose reaction preference towards the side chains and the aromatic core can be controlled by varying the reactions conditions (SSS-rule: sun, searing heat, side chain).

The crude phenanthrene from the previous step was dissolved in 40 mL of methanol, and

5 drops of elemental bromine were added. The mixture was boiled under reflux for 15 min, while in the presence of UV light (UVB, 2x15 W). The purified phenanthrene was recrystallized out of the reaction solution and further purified by two additional recrystallization steps from ethanol.

The final yield of phenanthrene was 4.24 g, which was free (<1 ppm) of any fluorene, anthracene or alkylated phenanthrene impurities.

9H-Fluorene

Commercial 9H-fluorene (98% Acros Organics, Morris Plains, NJ, USA) contained mainly methylated compounds, naphthalene and dibenzofuran as impurities. The methylated fluorenes are difficult to remove via chemical reaction (e.g. via bromination) as their reactivity is comparable with the methylene group of fluorene itself. For the removal of polar impurities such as dibenzofuran, 22.02 g of commercial 9H-fluorene was dissolved in 250 mL hexane and filtered through an alumina column (25 g). It was then subjected to five recrystallizations from ethanol/water to reduce the contribution of the alkylated impurities.

134 Acenaphthene

Crude acenaphthene (99% Acros Organics, Morris Plains, NJ, USA) was vacuum sublimed and recrystallized twice from dichloromethane.

Pyrene

The utilized pyrene (98%, scintillation grade, Sigma Aldrich, St. Louis, MO, USA) had methylpyrenes as its main impurities, which could not be removed by side-chain bromination, as pyrene itself reacted with the bromine too quickly. Therefore the commercial pyrene was recrystallized three times out of ethanol, to limit the effects of the impurities.

135 APPENDIX D: UV SPECTRAL MATCHES

For each compound identified in this thesis one representative comparison of the observed UV spectrum and the UV spectrum of a reference compound or a spectrum available in literature is given. The observed UV spectrum is marked as black line, whereas the reference compound is marked as red line.

Naphthalene Photooxidation Products

a) Naphthalene b) Naphthalene Photooxidation Product Photooxidation Product

Absorbance Absorbance

200 300 400 500 600 300 400 500 600 m/z m/z Figure 52. a) UV absorbance spectra of a naphthalene photooxidation product (black line) and phtalide (red line), and b) UV absorbance spectra of a naphthalene photooxidation product (black line) and coumarin (red line).

a) Naphthalene b) Naphthalene Photooxidation Product Photooxidation Product

Absorbance Absorbance

200 300 400 500 600 200 300 400 500 600 m/z m/z Figure 53. a) UV absorbance spectra of a naphthalene photooxidation product (black line) and 1- indanone (red line), and b) UV absorbance spectra of a naphthalene photooxidation product (black line) and 1,4-naphthoquinone (red line).

136 Naphthalene Photooxidation Product

Absorbance

200 300 400 500 600 m/z Figure 54. UV absorbance spectra of a naphthalene photooxidation product (black line) and naphthalene-1-ol (red line). Phenanthrene Photooxidation Products

a) Phenanthrene (impure) b) Photooxidation Product Phenanthrene Photooxidation Product

Absorbance

Absorbance

200 300 400 500 600 200 300 400 500 600 m/z m/z Figure 55. a) UV absorbance spectra of a phenanthrene photooxidation product (black line) and 9H-fluorene (red line), and b) UV absorbance spectra of a phenanthrene photooxidation product (black line) and 9H-fluorene-9-one (red line).

a) Phenanthrene b) Phenanthrene Photooxidation Product Photooxidation Product

Absorbance Absorbance

200 300 400 500 600 200 300 400 500 600 m/z m/z Figure 56. a) UV absorbance spectra of a phenanthrene photooxidation product (black line) and 6H-dibenzo[b,d]pyran-6-one (red line), and b) UV absorbance spectra of a phenanthrene photooxidation product (black line) and 9,10-phenanthrenequinone (red line).

137 Acenaphthene Photooxidation Products

a) Acenaphthene b) Acenaphthene Photooxidation Product Photooxidation Product

Absorbance Absorbance

200 300 400 500 600 200 300 400 500 600 m/z m/z Figure 57. a) UV absorbance spectra of a acenaphthene photooxidation product (black line) and 1-acenaphthenone (red line), and b) UV absorbance spectra of a acenaphthene photooxidation product (black line) and 1-acenaphthenol (red line).

Pyrene Photooxidation Products

a) Pyrene b) Pyrene Photooxidation Photooxidation Product Product

Absorbance

Absorbance

200 300 400 500 600 200 300 400 500 600 m/z m/z Figure 58. a) UV absorbance spectra of a pyrene photooxidation product (black line) and 1,8- pyrenequinone (red line) obtained from the literature,177 and b) UV absorbance spectra of a pyrene photooxidation product (black line) and 1,6-pyrenequinone (red line).

138 Pyrene Photooxidation Product

Absorbance

200 300 400 500 600 m/z Figure 59. UV absorbance spectra of a pyrene photooxidation product (black line) and 1- hydroxypyrene (red line)

9H-Fluorene Photooxidation Products

9H-Fluorene Photooxidation Product

Absorbance

200 300 400 500 600 m/z

Figure 60. UV absorbance spectra of a 9H-fluorene photooxidation product (black line) and 9H- fluorene-9-ol (red line)

139 APPENDIX E: SYNTHETIC PROCEDURES

This appendix describes the syntheses of the individual compounds identified during the course of this work. References, describing procedures as they were adapted from literature, can be found in the caption of the first figure describing each reaction.

Melting points (m.p.) in this section were obtained on a melting apparatus Isothermal

IA 9300 and are non-calibrated. One-dimensional 1H-nuclar magnetic resonance (NMR) spectrometry data were obtained on a Bruker VL 400 MHz NMR spectrometer with either dichloromethane-d2, chloroform-d3 or dimethylsulfoxide-d6 as solvent and tetramethylsilane

(TMS) as chemical shift reference standard at room temperature. Gas chromatographic analysis were conducted on a gas chromatograph coupled to mass spectrometer (GC-MS), namely an

Agilent 6890 GC coupled to a 5973 single quadrupole mass spectrometer with a 30-m×0.25- mm×0.25-µm×30-m HP-5MS column. The injection volume was 2 µL with an injector temperature of 280°C and a split ratio of 1:5. Helium was supplied as carrier gas at constant pressure of 9.14 psig providing a linear velocity of 40 cm/s. The temperature program of the column oven initiated at 40°C with a hold of 3 min and continued with a rise to 280°C within 60 min followed by a 30 min hold at this temperature. The MS transfer line was set at 300°C, while the mass spectrometer analyzed the ions generated via electron impact (70 eV; 230°C) within a scan range of m/z 50 to 550.

UV absorbance spectra and DAPPI mass spectra were recorded on an 1100 Agilent system with method D described in Appendix A.

140 Synthesis of 6H-Dibenzo[b,d]pyran-6-one

Figure 61. Bayer-Villiger Oxidation of 9H-fluorene-9-one to 6H-dibenzo[b,d]pyran-6-one according to Mehta et al.164

10.04 g (55.7 mmol) 9H-fluorene-9-one were dissolved in 200 mL glacial acetic acid.

170 mL hydrogen peroxide (30%, 1.66 mol) was added to the solution and the mixture heated in a water bath to 50°C. The conversion of the reactants was tested daily via GC-MS and the in literature reported yield of 80-85% could not be achieved within the reported 72 hours. After five days the reaction was therefore quenched with water and the yellowish white precipitate was filtered off by vacuum filtration. The raw mixture (8.08 g) was taken up by 12.89 g silica gel and separated on 115 g silica gel with a dichloromethane/hexane mixture (90:10) as mobile phase.

The first 250 mL contained pure 6H-dibenzo[b,d]pyran-6-one (1.55 g), whereas the following

160 mL contained a mixture of 6H-dibenzo[b,d]pyran-6-one and 9H-fluorenone (6.45 g).

The latter fraction was further separated in two batches by utilizing a linear hexane/dichloromethane gradient (100% hexane to 100% dichloromethane in 1 hr) at a flow rate of 5 mL/min on two commercial silica columns (Interchim PuriFlash 15 μm 25 g) on a flash chromatography system (Agilent 1200 quaternary pump G1311A and Advantec Super Fraction

Collector SF-2120). Each individual batch obtained from the 6H-dibenzo[b,d]pyran-6-one fraction was tested for purity by its melting point and the combined yield of these two fractions and the previously obtained pure 6H-dibenzo[b,d]pyran-6-one resulted in a total of 3.493 g (17.8

141 mmol) 6H-dibenzo[b,d]pyran-6-one with a combined, melting point of 93.5 to 94.5°C (Mehta et al.164 93-94°C). Figure 61 shows the 400 MHz proton NMR spectrum of 6H-dibenzo[b,d]pyran-

6-one in deuterated chloroform with tetramethylsilane as chemical shift reference standard. The

NMR analysis confirmed the identity of 6H-dibenzo[b,d]pyran-6-one with a total of eight protons in the aromatic region 7.3 to 8.5 ppm. Figure 62 shows the UV spectrum and the dopant

(benzene) assisted APPI mass spectrum recorded on the HPLC-UV/DAD-APPI/MS, which shows the main ion at m/z 197, with no ions found in the negative spectrum.

CDCl3

7.38, 1H, t

8.11, 1H, d 7.52, 1H, t 8.17, 1H, d 7.41, 1H, d 7.62, 1H, t

Intensity

8.45, 1H, d 7.87, 1H, t

8.6 8.4 8.2 8.0 7.8 7.6 7.4 7.2 7.0 Chemical Shift [ppm]

1 Figure 62. 400 MHz H-NMR spectrum of 6H-dibenzo[b,d]pyran-6-one in CDCl3

142 100 197 [M+H]+ 80

60

40

Absorbance 20

Relative Ion Abundance

0 200 300 400 500 600 160 180 200 220 240 Wavelength [nm] m/z

Figure 63. UV absorbance spectrum and mass spectrum of 6H-dibenzo[b,d]pyran-6-one.

Synthesis of 6H-Dibenzo[b,d]pyran-6-ol

Figure 64. Synthesis of 6H-dibenzo[b,d]pyran-6-ol following the procedure of Moliner et al.160

1.4559 g (7.42 mmol) 6H-dibenzo[b,d]pyran-6-one were dissolved in 50 mL dichloro- methane and cooled in a dry ice/acetone bath to -70°C, while keeping the solution blanketed under a nitrogen atmosphere. 12 mL diisobutylaluminium hydride (DIBAH) solution in DCM were added and the reaction mixture stirred for an hour, after which the reaction was slowly quenched with 10 mL water. The mixture was let stand to warm to room temperature and it was filtered after the addition of 1.2 g of celite. After separation and evaporation of the dichloromethane solution 1.1854 g of crude white 6H-dibenzo[b,d]pyran-6-ol was obtained.

143 Recrystallization from hexane/dichloromethane yielded 884.5 mg (4.46 mmol) 6H- dibenzo[b,d]pyran-6-ol with a melting point of 90°C.

Cyclohexane impurity

7.08-7.80, m TMS

CD Cl 3 3.10, d

Intensity 6.73, s 7.86, d 6.54, s

6.41, d

8 6 4 2 0 Chemical Shift [ppm]

1 Figure 65. 400 MHz H-NMR spectrum of crude 6H-dibenzo[b,d]pyran-6-one in CDCl3

The NMR in Figure 64 shows the combination of the spectra of 6H-dibenzo[b,d]pyran-6- ol (3.1 ppm, d; 6.54 ppm, s, 7.85 ppm, d),161 its dimer, 6,6’-oxybis-6H-dibenzo[b,d]pyran (6.73 ppm, 2H) and an unknown impurity (6.73 ppm, s). The UV spectrum and the corresponding mass spectrum of 6H-dibenzo[b,d]pyran-6-ol is shown in Figure 66.

144 100 181 [M-17]+ 80

60

40 Absorbance + 20 198 [M] Relative Ion Abundance 169 [M-29]- 197 [M-H]- 0 200 300 400 500 600 160 180 200 220 240 Wavelength [nm] m/z

Figure 66. UV absorbance spectrum and mass spectrum of 6H-dibenzo[b,d]pyran-6-ol

6H-dibenzo[b,d]pyran-6-ol has the tendency to react even as a solid to its dimer 6,6’- oxybis-6H-dibenzo[b,d]pyran. To recover the monomer it is necessary to boil the dimer in tetrahydrofuran/water with a catalytic amount of hydrochloric acid.

Synthesis of 2-(2’-Hydroxy-phenyl)-benzyl alcohol

Figure 67. Reduction of 6H-dibenzo[b,d]pyran-6-one with sodium borohydride resulting in 2- (2’-hydroxy-phenyl) benzyl alcohol

An initial attempt of reducing 6H-dibenzo[b,d]pyran-6-one to 6H-dibenzo[b,d]pyran-6-ol with sodium borohydride was conducted with tetrahydrofuran as solvent. Unfortunately, the reaction did not stop at the reduction of the ester (lacton) to the semi-acetal (lactol), but

145 continued further to the alcohol. The quenching reaction with water was quite violent and resulted in the ejection of large amounts of reaction mixture through the utilized Liebig cooler, resulting in diminished yield. Although this compound has not been observed as product of fluorene or phenanthrene photo-oxidation, it is reported here for the sake of completeness, as its synthesis via this procedure (reduction with NaBH4) seems to be a more promising avenue than the in literature described procedure, which resulted in a more impure, brown-colored product.178

For future reproduction it is highly recommended to initiate the quenching step very slowly after the solution has been pre-cooled with ice.

205.3 mg (1.047 mmol) 6H-dibenzo[b,d]pyran-6-one was dissolved in 20 mL tetrahydrofuran and 403 mg (10.65 mmol) sodium borohydride was added. The mixture was refluxed for 2.5 hours. After letting the mixture cool partially, water was added, upon which the mixture started to boil violently, ejecting material through the condenser. After subsiding of the violent reaction the mixture was completely quenched with more water and extracted with dichloromethane. Evaporation of the dichloromethane yielded an oily liquid which crystallized from cyclohexane/dichloromethane as white mass. HPLC-UV/DAD-DAPPI/MS showed that the mixture contained mainly 2-(2’-hydroxy-phenyl)-benzyl alcohol with small parts of unreacted

6H-dibenzo[b,d]pyran-6-one.

The NMR spectrum of the crude 2-(2’-hydroxy-phenyl)-benzyl alcohol (the NMR spectrum of the 6H-dibenzo[b,d]pyran-6-one impurity is limited to the aromatic region and therefore not hindering the analysis) is shown in Figure 67. Apart from the protons in the aromatic region of 7.01-7.6 ppm, there are three more proton types visible. The integration of the peak at 4.53 ppm amounting to two protons, allows its assignment to the methylene group. By comparing the shift values of the current compound with those of 4-phenylbenzyl alcohol179

146 (methylene group 4.687 ppm and hydroxy group 2.1 ppm) it is evident that the proton signal at

2.1 ppm belongs to the hydroxy-proton connected to the benzyl group. The peak at 5.7 ppm can be assigned based on exclusion to the phenolic hydroxy group, which is also consistent with the comparable shift value for the hydroxy group for biphenyl-2-ol at 5.23 ppm as reported in literature.179 The UV spectrum and the corresponding mass spectrum of 2-(2’-hydroxy-phenyl)- benzyl alcohol is shown in Figure 69.

CDCl3 TMS

4.53, 2H, d Cyclohexane impurity 7.01-7.59, 8H, m

Intensity

5.70, 1H, d 2.10, 1H, s

8 6 4 2 0 Chemical Shift [ppm]

Figure 68. 400 MHz 1H-NMR spectrum of 2-(2’-hydroxy-phenyl)-benzyl alcohol in CDCl3

147 100 181 [M-18]+ 80 182 [M-19]+

60

40

Absorbance 199 [M-H]- 20

Relative Ion Abundance 169 [M-31]-

0 200 300 400 500 600 160 180 200 220 240 Wavelength [nm] m/z

Figure 69. UV absorbance spectrum and mass spectrum of 2-(2’-hydroxyphenyl)-benzyl alcohol

Synthesis of 2- and 3- Hydroxy-9,10-phenanthrenequinone

Figure 70. Synthetic scheme for the synthesis of 2- and 3-hydroxy-9,10-phenanthrenequinone according to Fieser180 and Werner et al.148-149

2-hydroxy- and 3-hydroxy-9,10-phenanthrenequinone were synthesized via a multistep procedure shown in Figure 69. At first the 2- and 3-phenanthrene sulfonic acids are generated

148 following the procedure of Fieser,180 with separation of the isomers via their barium and potassium salts. Subsequently the sulfonates are transformed via alkali melt into the respective hydroxy derivatives.181 These phenolic compounds are esterified to their acetoxy derivatives to protect the hydroxy group during the following oxidation with chromium trioxide. Upon removal of the protecting acetyl group, 2- and 3-hydroxy-9,10-phenanthrenequinone can be obtained in pure form.

Synthesis of 2- and 3-Phenanthrene Sulfonates

Figure 71. Synthesis of 2- and 3-phenanthrene sulfonate according to Fieser180

The hydroxyphenanthrenes were synthesized from the sulfonic acids by aryl sulfonation of phenanthrene and separation of the individual isomers based on the different solubility of their the barium and potassium salts following the procedure of Fieser.180

20.047 g phenanthrene (112.48 mmol) was made molten in a 3-neck flask in an oil bath heated to 110°C. 25.4434 g (254.23 mmol) concentrated sulfuric acid (98%) was added within

16 minutes, while keeping the temperaure below 125°C. The mixture was stirred for 3.5 hours, after which it was poured into 320 mL water and 30 ml of 35% sodium hydroxide solution. The mixture was cooled in an ice bath and the precipitate harvested via filtration. After pressing the solids dry, they were washed with half-saturated sodium chloride solution.

149 The precipitate was redissolved in 320 mL water to which 4 mL concentrated hydrochloric acid was added. The solution was brought to a boil and neutralized with sodium hydroxide while boiling. Upon cooling a white precipitate could be observed. The solids were separated via filtration and redissolved in 320 mL boiling water to which 4 g barium chloride dihydrate salt was added. The resulting fine precipitate was separated after cooling. The crude barium phenanthrene-2-sulfonate was extracted twice with hot boiling water to remove the 3- isomer, yielding 8.851 g (22.4 mmol) 2-barium phenanthrene sulfonate.

The hot-water extracts, as well as the filtrate of the previous barium precipitation were combined and the barium removed by addition of concentrated sulfuric acid and filtration of the precipitated barium sulfate. The filtered solution was boiled down to approximately 80 mL and neutralized with potassium hydroxide. After cooling, a crop of white crystals of potassium phenanthrene-3-sulfonate could be obtained. These crystals were digested in 80 mL of boiling water and as the addition of barium chloride did not yield any precipitate of the 2-isomer let stand to crystallize. A total of 1.631 g of potassium phenanthrene-3-sulfonate was obtained.

Synthesis of 2-Hydroxyphenanthrene

Figure 72. Transformation of barium 2-phenanthrenesulfonate into 2-hydroxyphenanthrene181

15.9 g potassium hydroxide and 4.997 g barium phenanthrene-2-sulfonate are made molten in a iron crucible and kept molten for a few min. The hot reaction mixture was poured

150 into a large beaker and dissolved with 175 mL water. Neutralization with 27 mL concentrated hydrochloric acid yielded an organic precipitate, which was extracted with dichloromethane and the solvent of the extract evaporated on a rotary evaporator. The resulting brown residue was taken up again in dichloromethane and filtered through a silica column. The eluting mixture was boiled with active carbon for a rather unsuccessful decolorization and recrystallized from dichloromethane/hexanes yielding 571.1 mg of 2-hydroxyphenanthrene. The NMR spectrum of

2-hydroxyphenanthrene is shown in Figure 73 and is congruent with the spectrum available in literature.179

7.22-7.87, m TMS

CDCl3 8.59, 2H, ss

5.04, 1H, s Hexane

Intensity impurity

8 6 4 2 0 Chemical Shift [ppm]

1 Figure 73. 400 MHz H-NMR spectrum of 2-hydroxyphenanthrene in CDCl3

151 The UV spectrum and the corresponding mass spectrum of 2-hydroxyphenanthrene is shown in

Figure 74.

100 193 [M-H]- 194 [M]+

80

60

40

Absorbance 20

Relative Ion Abundance

0 200 300 400 500 600 160 180 200 220 240 Wavelength [nm] m/z

Figure 74. UV absorbance spectrum and mass spectrum of 2-hydroxyphenanthrene

Synthesis of 2-Acetoxyphenanthrene and 2-Acetoxy-9,10- phenanthrenequinone

Figure 75. Acetylation of 2-hydroxyphenanthrene and oxidation182 to 2-acetoxy-9,10- phenanthrenequinone

152 423.2 mg 2-hydroxyphenanthrene was refluxed with 5 mL acetic acid anhydride and 10 mL acetic acid for 7.5 hours. After the solution was cooled 845.9 mg chromium trioxide was added within 15 minutes in an attempt to combine the acetylation and oxidation step. The reaction was quenched with 100 mL water and the mixture extracted with dichloromethane.

After evaporation of the combined extracts 406.9 mg material was obtained. The conversion of the reaction was checked via GC-MS and only approximately half of the acetoxyphenanthrene had been converted to the corresponding quinone. As such the oxidation was conducted a second time by dissolving the material in acetic acid and adding this solution to 407.5 mg chromiumtrioxide dissolved in 10 mL 80% acetic acid within 15 min. The solution was continuously stirred for 2 hours at 45°C. The reaction was quenched by pouring the reaction mixture into 150 mL water mixed with 150 mL saturated sodium hydrogencarbonate solution.

The oxidized acetoxyphenanthrenequinone was extracted with dichloromethane. After removal of the solvent in the rotary evaporator 282.4 mg of 2-acetoxy-9,10-phenanthrenequinone was found.

Synthesis of 2-Hydroxy-9,10-phenanthrenequinone

Figure 76. Hydrolysis of 2-acetoxy-9,10-phenanthrenequinone to 2-hydroxyphenanthrene- quinone

153 282.4 mg (1.06 mmol) 2-acetoxy-9,10-phenanthrenequinone were mixed with 23 mL

23% sodium hydroxide solution and 50 mL water upon which a deep-red solution formed. By acidifiying the solution with 3 mL concentrated hydrochloric acid a dark colored precipitate could be obtained. After recrystallization out of water/ethanol 125.5 mg of black-purple crystals without observable melting point could be obtained. The NMR spectrum of 2-hydroxy-9,10- phenanthrenequinone is shown in Figure 77 and its UV an mass spectrum are shown in Figure

78.

7.11-8.06, 7H, m H2O

10.20, 1H, s DMSO-d6

Intensity

TMS

12 10 8 6 4 2 0 Chemical Shift [ppm]

Figure 77. 400 MHz 1H-NMR spectrum of 2-hydroxy-9,10-phenanthrenequinone in DMSO-d6

154 100 225 [M+H]+

80

60

40 Absorbance 223 [M-H]- 20

Relative Ion Abundance 197 [M+H-CO]+

0 200 300 400 500 600 160 180 200 220 240 Wavelength [nm] m/z

Figure 78. UV absorbance spectrum and mass spectrum of 2-hydroxy-9,10- phenanthrenequinone

Synthesis of 3-Hydroxyphenanthrene

Figure 79. Transformation of potassium phenanthrene-3-sulfonate into 3- hydroxyphenanthrene181

1631.0 mg potassium phenanthrene-3-sulfonate and 7.1 g potassium hydroxide were made molten in a beaker. The 3-hydroxyphenanthrene formed an oily layer on top of the molten alkali.

After a few minutes the melt was let cool down and the hardened mixture taken up into 300 mL water. The addition of 30 mL concentrated hydrochloric acid caused the 3-hydroxyphenanthrene to precipitate. The crude brown solids (890.8 mg) were taken up by dichloromethane and filtered through a silica gel column (1 cm diameter, 5.98 g silica gel). Collection of the first 120 mL and evaporation of this fraction allowed the collection of 830.5 mg

155

7.16-8.04, m

8.53, 1H, s TMS CDCl3

5.13, 1H, s Hexane

Intensity impurity

8 6 4 2 0 Chemical Shift [ppm]

1 Figure 80. 400 MHz H-NMR spectrum of 3-hydroxyphenanthrene in CDCl3

100 193 [M-H]-

80 194 [M]+ 60

40

Absorbance 20

Relative Ion Abundance

0 160 180 200 220 240 200 300 400 500 600 m/z Wavelength [nm]

Figure 81. UV absorbance spectrum and mass spectrum of 3-hydroxyphenanthrene

156 crude 3-hydroxyphenanthrene with a melting point range of 112.8 to 117.2°C. The NMR spectrum of 2-hydroxyphenanthrene is shown in Figure 80 and is congruent with the spectrum available in literature.179 Figure 81 depicts the UV spectrum and the APPI-MS spectrum of 3- hydroxyphenanthrene.

Synthesis of 3-Acetoxyphenanthrene

Figure 82. Acetylation of 3-hydroxyphenanthrene to 3-acetoxyphenanthrene

400 mg (2.06 mmol) 3-hydroxyphenanthrene were dissolved in 20 mL glacial acetic acid and 5 ml (53.4 mmol) acetic anhydride were addded. The mixture was refluxed for four hours.

Upon addition of 40 mL deionized water the 3-acetoxyphenanthrene precipitated and was recovered via vacuum filtration. After drying 428.2 mg (1.81 mmol) of 3-acetoxyphenanthrene were obtained. The purity and as such the extent of conversion was checked with GC-MS (E.I.; m/z 194, 165, 236). The sample was of sufficient purity (m.p. 109.6-116.8°C) to allow the use of the crude compound for the next reaction step without further purification, especially as the main impurity is unreacted 3-hydroxyphenanthrene, which is easily removed by washing the very polar hydroxyphenanthrenequinone (final product) with dichloromethane (non-polar) in the final step.

157 Synthesis of 3-Acetoxy-9,10-phenanthrenequinone

Figure 83. Oxidation of 3-acetoxyphenanthrene to 3-acetoxyphenanthrene-9,10-quinone with chromium trioxide182

Following loosely the procedure of Alarcón et al.182 428.2 mg (1.81 mmol) 3- acetoxyphenanthrene were dissolved in 17 mL acetic acid and over 4 hours slowly added to an ice-bath-cooled solution of 454.8 mg (4.54 mmol) chromium trioxide in 11 mL 80% acetic acid.

The solution was let stand for 24 hours. Upon addition of 80 mL water a canary-yellow precipitate could be obtained from the solution by filtration.

After drying a total of 200.2 mg (0.75 mmol) of 3-acetoxy-9,10-phenanthrenequinone

(m.p. 177-185.7°C w/decomposition GC-MS; EI m/z 194, 100%; m/z 165, 43%; m/z 236, 15%) could be obtained.

158 Synthesis of 3-Hydroxy-9,10-phenanthrenequinone

Figure 84. Hydrolysis of 3-acetoxy-9,10-phenanthrenequinone to 3- hydroxyphenanthrenequinone

The entire crop (200.2 mg, 0.75 mmol) of the previously obtained 3-acetoxyphenanthrene-9,10- quinone was taken up in 20 mL dichloromethane and extracted three times with a total of 100 mL water to which 5 mL 20% sodium hydroxide had been added. The aqueous solutions were separated and acidified with concentrated hydrochloric acid. The resulting orange precipitate was filtered off and washed thrice with 10 mL dichloromethane. The dichloromethane solutions

(wash and extracted solution) were evaporated and the residue dissolved in 100 mL dilute sodium hydroxide solution while being heated on a water bath. The deep red solution was again acidified and the orange precipitate filtered off and washed thrice with dichloromethane. The dried brick-red material was dissolved in 160 mL ethanol and recrystallized in two steps to yield a combined total of 177.6 mg (0.79 mmol) 3-hydroxy-9,10-phenanthrenequinone. The NMR spectrum of 3-hydroxy-9,10-phenanthrenequinone is shown in Figure 85. The UV spectrum and the corresponding mass spectrum obtained from HPLC-DAD/UV-APPI/MS analysis of 3- hydroxy-9,10-phenanthrenequinone is shown in Figure 86.

159 6.89-8.00, 7H, m H2O

DMSO-d6

11.01, 1H, s

Intensity

TMS

12 10 8 6 4 2 0 Chemical Shift [ppm]

Figure 85. 400 MHz 1H-NMR spectrum of 3-hydroxyphenanthrene-9,10-quinone in DMSO-d6

100 225 [M+H]+

80

60

40

Absorbance 223 [M-H]- 20

Relative Ion Abundance 197 [M+H-CO]+

0 200 300 400 500 600 160 180 200 220 240 Wavelength [nm] m/z

Figure 86. UV absorbance spectrum and mass spectrum of 3-hydroxyphenanthrene-9,10- quinone

160 Synthesis of 1-Acenaphthenone

Figure 87. Oxidation of acenaphthene with chromium trioxide

3 g acenaphthene were dissolved in 50 ml acetic acid with 5 ml acetic acid anhydride in a

125-mL round bottom flask in an ice bath. 2.6 g chromium trioxide dissolved in 50 mL 80% acetic acid was added slowly over a time period of 3.6 hours, while the temperature was kept at

5-10°C. The solution was stirred over night and the reaction stopped by addition of 750 mL water and 50 mL 20% sodium hydroxide solution. After neutralization the mixture was extracted with dichloromethane. After evaporation of the solvent 2.83 g crude material was obtained. The mixture was taken up in dichloromethane and deposited onto 5 g silica gel. The crude product was purified via chromatography on 25 g silica with 90% hexane 10% DCM as mobile phase yielding 1.77 g material. As the purity of this product was still insufficient (tested with GC-MS) the mixture was again chromatographed on a silica column with a gradient run starting at a mobile phase composition of 90% dichloromethane/10% hexane ramping up to pure dichloro- methane within one hour. Fractions were collected every two minutes at a flow rate of 5 mL/min.

The first 160 mL contained mainly acenaphthene, while the following 100 mL contained acenaphthenone. After evaporation of the solvent 798.4 mg acenaphthenone with a melting point range of 120.7 to 122.8°C was obtained. The NMR spectrum of acenaphthenone is shown in

Figure 88 and its UV and APPI-mass spectrum is depicted in Figure 89.

161 DCM impurity

3.83, 2H, s

7.47-7.72, 6H, m

Intensity TMS CDCl3

10 8 6 4 2 0 Chemical Shift [ppm]

1 Figure 88. 400 MHz H-NMR spectrum of 1-acenaphthenone in CDCl3

100 168 [M]+ + 80 169 [M+H]

60

40

Absorbance - 20 167 [M-H]

Relative Ion Abundance

0 200 300 400 500 600 150 160 170 180 190 200 Wavelength [nm] m/z

Figure 89. UV absorbance spectrum and mass spectrum of 1-acenaphthenone

162 APPENDIX F: PERMISSION TO REPRINT

163 VITA

Franz Stefan Ehrenhauser was born in Linz, Austria, in November 1978. In 1999, he started his academic education at the Johannes Kepler University, Linz, Austria, studying

Wirtschaftsingenieurwesen Technische Chemie (Technical Chemistry / Economics). After going abroad in 2003 for four month as intern in Nakskov, Denmark, he continued his international experience by enrolling at Louisiana State University in Fall 2004 as exchange student. In

Summer 2005 he commenced as a graduate student to work on the identification of photooxidation products of polycyclic aromatic hydrocarbons under the guidance of Dr. M. J.

Wornat. While fully enrolled at LSU, he completed in July 2009 his diploma studies (Diplom

Ingenieur) in Austria, with his thesis titled ―Solubility behavior of natural substances in sub- and supercritical carbon dioxide.‖ Franz Stefan Ehrenhauser is co-author of four collaborative publications and he is the first author in two publications. He has given ten presentations at various conferences and was awarded the ―Best Student Presentation Award‖ at the 22nd

International Symposium on Polycyclic Aromatic Compounds (ISPAC). From 2009 to 2011

Franz Stefan Ehrenhauser received the Coates Scholar Research Grant from LSU. He is expected to graduate in Fall 2011 with the degree Doctor of Philosophy in chemical engineering.

164